Professional Documents
Culture Documents
Eduardo W. V. Chaves (Auth.) Notes On Continuum Mechanics 2013 PDF
Eduardo W. V. Chaves (Auth.) Notes On Continuum Mechanics 2013 PDF
on Continuum Mechanics
Eduardo W.V. Chaves
Lecture Notes
on Numerical Methods
in Engineering and Sciences
Series Editor
Eugenio Oate
International Center for Numerical Methods in Engineering (CIMNE)
School of Civil Engineering, Technical University of Catalonia (UPC), Barcelona, Spain
Editorial Board
Francisco Chinesta, Ecole Nationale Suprieure d'Arts et Mtiers, Paris, France
Charbel Farhat, Stanford University, Stanford, USA
Carlos Felippa, University of Colorado at Boulder, Colorado, USA
Antonio Huerta, Technical University of Catalonia (UPC), Barcelona, Spain
Thomas J.R. Hughes, The University of Texas at Austin, Austin, USA
Sergio R. Idelsohn, CIMNE-ICREA, Barcelona, Spain
Pierre Ladeveze, ENS de Cachan-LMT-Cachan, France
Wing Kam Liu, Northwestern University, Evanston, USA
Xavier Oliver, Technical University of Catalonia (UPC), Barcelona, Spain
Manolis Papadrakakis, National Technical University of Athens, Greece
Jacques Priaux, CIMNE-UPC Barcelona, Spain & Univ. of Jyvskyl, Finland
Bernhard Schrefler, Universit degli Studi di Padova, Padova, Italy
Genki Yagawa, Tokyo University, Tokyo, Japan
Mingwu Yuan, Peking University, China
Titles:
1. E. Oate, Structural Analysis with the Finite Element Method.
Linear Statics. Volume 1. Basis and Solids, 2009
2. K. Winiewski, Finite Rotation Shells. Basic Equations and
Finite Elements for Reissner Kinematics, 2010
3. E. Oate, Structural Analysis with the Finite Element Method.
Linear Statics. Volume 2. Beams, Plates and Shells, 2013
4. E.W.V. Chaves. Notes on Continuum Mechanics. 2013
To my Parents
Contents
Contents
PREFACE .......................................................................................................................................................XIX
ABBREVIATIONS .......................................................................................................................................... XXI
OPERATORS AND SYMBOLS....................................................................................................................XXIII
SI-UNITS ..................................................................................................................................................... XXV
INTRODUCTION ................................................................................................................. 1
1 MECHANICS...............................................................................................................................................1
2 WHAT IS CONTINUUM MECHANICS?....................................................................................................1
2.1 Hypothesis of Continuum Mechanics .........................................................................................1
2.2 The Continuum ...............................................................................................................................2
3 SCALES OF MATERIAL STUDIES.............................................................................................................3
3.1 Scale Study of Continuum Mechanics .........................................................................................3
4 THE INITIAL BOUNDARY VALUE PROBLEM (IBVP) .........................................................................6
4.1 Solving the IBVP.............................................................................................................................6
4.2 Simplifying the IBVP......................................................................................................................7
1 TENSORS.............................................................................................................................9
1.1 INTRODUCTION.....................................................................................................................................9
1.2 ALGEBRAIC OPERATIONS WITH VECTORS ....................................................................................10
1.3 COORDINATE SYSTEMS .....................................................................................................................16
1.3.1 Cartesian Coordinate System....................................................................................................16
1.3.2 Vector Representation in the Cartesian Coordinate System ...............................................17
1.3.3 Einstein Summation Convention (Einstein Notation) ........................................................20
1.4 INDICIAL NOTATION .........................................................................................................................20
1.4.1 Some Operators..........................................................................................................................22
1.4.1.1 Kronecker Delta.............................................................................................................22
1.4.1.2 Permutation Symbol ......................................................................................................23
1.5 ALGEBRAIC OPERATIONS WITH TENSORS.....................................................................................28
1.5.1 Dyadic ..........................................................................................................................................28
1.5.1.1 Component Representation of a Second-Order Tensor in the Cartesian
Basis...................................................................................................................................32
1.5.2 Properties of Tensors ................................................................................................................34
1.5.2.1 Tensor Transpose ..........................................................................................................34
1.5.2.2 Symmetry and Antisymmetry.......................................................................................36
1.5.2.3 Cofactor Tensor. Adjugate of a Tensor .....................................................................42
1.5.2.4 Tensor Trace...................................................................................................................42
1.5.2.5 Particular Tensors ..........................................................................................................44
1.5.2.6 Determinant of a Tensor ..............................................................................................45
1.5.2.7 Inverse of a Tensor........................................................................................................48
1.5.2.8 Orthogonal Tensors ......................................................................................................51
VII
VIII
CONTENTS
IX
2 CONTINUUM KINEMATICS........................................................................................145
2.1 INTRODUCTION.................................................................................................................................145
2.2 THE CONTINUOUS MEDIUM ..........................................................................................................146
2.2.1 Kinds of Motion.......................................................................................................................147
2.2.1.1 Rigid Body Motion ......................................................................................................147
2.2.2 Types of Configurations .........................................................................................................149
2.2.2.1 Mass Density.................................................................................................................150
2.3 DESCRIPTION OF MOTION .............................................................................................................151
2.3.1 Material and Spatial Coordinates ...........................................................................................151
2.3.2 The Displacement Vector.......................................................................................................152
2.3.3 The Velocity Vector.................................................................................................................152
2.3.4 The Acceleration Vector .........................................................................................................152
2.3.5 Lagrangian and Eulerian Descriptions..................................................................................152
2.3.5.1 Lagrangian Description of Motion............................................................................152
2.3.5.2 Eulerian Description of Motion ................................................................................153
2.3.5.3 Lagrangian and Eulerian Variables............................................................................153
2.4 THE MATERIAL TIME DERIVATIVE ..............................................................................................156
2.4.1 Velocity and Acceleration in Eulerian Description ............................................................158
2.4.2 Stationary Fields .......................................................................................................................159
2.4.3 Streamlines ................................................................................................................................161
2.5 THE DEFORMATION GRADIENT ...................................................................................................163
2.5.1 Introduction ..............................................................................................................................163
2.5.2 Stretch and Unit Extension ....................................................................................................163
2.5.3 The Material and Spatial Deformation Gradient ................................................................165
2.5.4 Displacement Gradient Tensors (Material and Spatial) .....................................................168
2.5.5 Material Time Derivative of the Deformation Gradient. Material Time Derivative
of the Jacobian Determinant................................................................................................171
2.5.5.1 Material Time Derivative of F . The Spatial Velocity Gradient ..........................171
2.5.5.2 Rate-of-Deformation and Spin Tensors...................................................................172
2.5.5.3 The Material Time Derivative of F 1 ......................................................................174
2.5.5.4 The Material Time Derivative of the Jacobian Determinant ................................174
2.6 FINITE STRAIN TENSORS.................................................................................................................176
2.6.1 The Material Finite Strain Tensor .........................................................................................177
2.6.2 The Spatial Finite Strain Tensor (The Almansi Strain Tensor) ........................................181
2.6.3 The Material Time Derivative of Strain Tensors ................................................................183
2.6.3.1 The Material Time Derivative of the Right Cauchy-Green Deformation
Tensor .............................................................................................................................183
2.6.3.2 The Material Time Derivative of the Green-Lagrange Strain Tensor .................183
2.6.3.3 The Material Time Derivative of C 1 ......................................................................184
2.6.3.4 Material Time Derivative of the Left Cauchy-Green Deformation Tensor.......184
2.6.3.5 The Material Time Derivative of the Almansi Strain Tensor ...............................185
2.6.4 Interpreting Deformation/Strain Tensors ...........................................................................186
2.6.4.1 The Relationship between the Strain and Stretch Tensors ...................................187
2.6.4.2 Change of Angle...........................................................................................................188
2.6.4.3 The Physical Interpretation of the Deformation/Strain Tensor
Components. The Right Stretch Tensor ...................................................................189
2.7 PARTICULAR CASES OF MOTION ...................................................................................................191
2.7.1 Homogeneous Deformation ..................................................................................................191
2.7.2 Rigid Body Motion...................................................................................................................192
2.8 POLAR DECOMPOSITION OF F .....................................................................................................195
2.8.1 Spectral Representation of Kinematic Tensors...................................................................197
2.8.2 Evolution of the Polar Decomposition................................................................................203
2.8.2.1 The Alternative Way to Express the Rate of Kinematic Tensors........................208
2.9 AREA AND VOLUME ELEMENTS DEFORMATION ......................................................................215
2.9.1 Area Element Deformation....................................................................................................215
2.9.1.1 The Material Time Derivative of the Area Element .............................................. 217
2.9.2 The Volume Element Deformation ..................................................................................... 218
2.9.2.1 The Material Time Derivative of the Volume Element ........................................ 219
2.9.2.2 Dilatation....................................................................................................................... 220
2.9.2.3 Isochoric Motion. Incompressibility ........................................................................ 220
2.10 MATERIAL AND CONTROL DOMAINS ........................................................................................ 220
2.10.1 The Material Domain............................................................................................................ 220
2.10.2 The Control Domain ............................................................................................................ 221
2.11 TRANSPORT EQUATIONS .............................................................................................................. 222
2.12 CIRCULATION AND VORTICITY ................................................................................................... 224
2.13 MOTION DECOMPOSITION: VOLUMETRIC AND ISOCHORIC MOTIONS .............................. 225
2.13.1 The Principal Invariants ....................................................................................................... 227
2.14 THE SMALL DEFORMATION REGIME ......................................................................................... 228
2.14.1 Introduction............................................................................................................................ 228
2.14.2 Infinitesimal Strain and Spin Tensors ................................................................................ 229
2.14.3 Stretch and Unit Extension.................................................................................................. 231
2.14.4 Change of Angle .................................................................................................................... 232
2.14.5 The Physical Interpretation of the Infinitesimal Strain Tensor ..................................... 232
2.14.5.1 Engineering Strain ..................................................................................................... 233
2.14.6 The Volume Ratio (Dilatation)............................................................................................ 235
2.14.7 The Plane Strain..................................................................................................................... 236
2.15 OTHER WAYS TO DEFINE STRAIN .............................................................................................. 239
2.15.1 Motivation............................................................................................................................... 239
2.15.2 The Logarithmic Strain Tensor ...........................................................................................241
2.15.3 The Biot Strain Tensor ......................................................................................................... 242
2.15.4 Unifying the Strain Tensors ................................................................................................. 242
2.15.5 One Dimensional Measurements of Strain (1D).............................................................. 243
2.15.5.1 Cauchys strain or Engineering strain or the Linear strain.................................. 243
2.15.5.2 The Logarithmic or True strain............................................................................... 243
2.15.5.3 The Green-Lagrange strain ...................................................................................... 243
2.15.5.4 The Almansi strain ....................................................................................................243
2.15.5.5 The Swaiger strain ..................................................................................................... 244
2.15.5.6 The Kuhn strain......................................................................................................... 244
3 STRESS .............................................................................................................................245
3.1 INTRODUCTION ................................................................................................................................ 245
3.2 FORCES ............................................................................................................................................... 245
3.2.1 Surface Forces (Traction) ....................................................................................................... 245
3.2.2 Gravitational Force (Body Force) .........................................................................................246
3.3 STRESS TENSORS ...............................................................................................................................247
3.3.1 The Cauchy Stress Tensor...................................................................................................... 248
3.3.1.1 The Traction Vector....................................................................................................248
3.3.1.2 Cauchys Fundamental Postulate .............................................................................. 248
3.3.2 The Relationship between the Traction and the Cauchy Stress Tensor ......................... 252
3.3.3 Other Measures of Stress ....................................................................................................... 260
3.3.3.1 The First Piola-Kirchhoff Stress Tensor ................................................................. 260
3.3.3.2 The Kirchhoff Stress Tensor ..................................................................................... 262
3.3.3.3 The Second Piola-Kirchhoff Stress Tensor............................................................. 262
3.3.3.4 The Biot Stress Tensor ............................................................................................... 264
3.3.3.5 The Mandel Stress Tensor.......................................................................................... 264
3.3.4 Spectral Representation of the Stress Tensors.................................................................... 265
4 OBJECTIVITY OF TENSORS ........................................................................................269
4.1 INTRODUCTION ................................................................................................................................ 269
CONTENTS
XI
XII
CONTENTS
XIII
XIV
CONTENTS
XV
XVI
CONTENTS
XVII
Preface
Preface
The Continuum Mechanics is a key subject to several degrees based on physical science,
such as: Civil Engineering, Industrial Engineering, Meteorology, Magnetism,
Oceanography, Aerodynamics, Hydrodynamics, Marine Engineering, etc.
This book grew out of notes for the course Introduction to Continuum Mechanics of the
career of Civil Engineering of the University of Castilla-La Mancha (Spain), and is intended
for students who are initiating a university degree based on physical science, and is also
intended for PhD students as well researchers.
In order to provide greater clarity for students, this book presents a thorough detail at the
time of the demonstration of the equations. At the time of writing the book, the author has
had a big concern for trying to unify the existing nomenclature, and to this end has
consulted numerous articles and books on the subject. With respect to the notation, the
developments of the equations are indiscriminately presented in tensorial, inditial and Voigt
notations. Another aspect is that the book is self-contained, so that the concepts used are
defined in the text.
Finally, I would like to express my gratitude to: Houzeaux (Guillaume), Vzquez (Mariano),
Gallego (Inmaculada), Pulido (Loli), Bentez (Jos Mara), Casati (Mara Jesus), Vlez
(Eduardo), Solares (Cristina), Olivares (Miguel ngel), Escobedo (Fernando), Simarro
(Gonzalo), Sanz (Ana), for aid to the revision of the first edition in Spanish. I would also
like to thank Toby Wakely for reviewing the English.
I would also to thank two Professors who marked my teaching and research career: Prof.
Xavier Oliver and Prof. Wilson Venturini (in memoriam).
Eduardo W. V. Chaves
Ciudad Real-Spain, October 2012
XIX
Abbreviations
Abbreviations
IBVP
BVP
FEM
BEM
FDM
Latin
i.e.
et al.
e.g.
etc.
v., vs.
viz.
id est
et alii
exempli gratia
et cetera
versus
vidilicet
that is
and the others
for example
and so on
versus
namely
XXI
x x
2
x
Tr (x)
Macaulay bracket
(x) sph
Euclidian norm of x
trace of (x)
transpose of (x)
inverse of (x)
inverse of the transpose of (x)
symmetric part of (x)
antisymmetric (skew-symmetric) part of (x)
spherical part of (x)
(x) dev
module of x
(x) T
(x) 1
(x) T
(x) sym
(x) skew
>>x@@
detx { x
Dx
{ x
Dt
cof (x)
Adjx
Tr x
:
2
x { grad(x)
x { div (x)
jump of x
scalar product
determinant of x
material time derivative of x
cofactor of x ;
adjugate of x
trace of x
double scalar product (or double contraction or double dot product)
Scalar differential operator
tensorial product
gradient of x
divergence of x
vector product (or cross product)
XXIII
SI-Units
SI-Units
length
m - metro
mass
kg - kilogram
power
time
temperature
s - second
K - Kelvin
m
s
m
s2
velocity
acceleration
energy
force
N - Newton
pressure, stress
Pa {
Prefix
pico
nano
micro
mili
centi
deci
m
c
d
m2
dynamic viscosity
Pa u s
kg
m2s
J
m2s
W
mK
kg
m3
energy flux
thermal conductivity
N
- Pascal
m2
p
K
permeability
mass flux
Nm - Joules
Symbol
J Nm - Joules
J
{ W watt
s
mass density
Prefix
10 n
10
12
10 9
10
6
10 3
kilo
Mega
Giga
Tera
Symbol
10 n
10 3
M
G
10 6
1012
10 9
2
10
10
XXV
Introduction
Introduction
1 Mechanics
Broadly speaking, Mechanics is the branch of physics that studies the behavior of a body
when it is subjected to forces, (e.g. deformation) and how it evolves over time. In general,
Mechanics can be classified into:
Theoretical Mechanics;
Applied Mechanics;
Computational Mechanics.
Continuum Mechanics
Theoretical Mechanics establishes the laws that govern a particular physical problem based on
fundamental principles.
Applied Mechanics transfers theoretical knowledge to use it in scientific and engineering
problems.
Computational Mechanics solves problems by simulation with numerical tools implemented in
the computer.
In this book we focus our attention to the Theoretical and Applied Mechanics.
2.1
As we know, a physical body consists of small molecules (an agglomeration of two or more
atoms). Then, by means of sophisticated experiments, we can observe that these
constituents are not distributed homogeneously, that is, there are gaps (voids) between
them. However, within the scope of Continuum Mechanics these phenomenological
E.W.V. Chaves, Notes on Continuum Mechanics, Lecture Notes on Numerical
Methods in Engineering and Sciences 4, DOI 10.1007/978-94-007-5986-2_1,
International Center for Numerical Methods in Engineering (CIMNE), 2013
characteristics are ignored. For example, if we are dealing with a fluid in Continuum
Mechanics, the properties: mass density, pressure and velocity are assumed to be
continuous function. Treating a system of molecules as a continuous medium is valid if we
compare the mean free path of molecules ( / ) (average distance particles travel before
colliding with each other) with the characteristic physical length scale ( " C ). For example,
for solids and liquids we have / | 10 7 cm and for gases / | 10 6 cm , Chung (1996). Then
the ratio
"C
unity, the domain can be treated as a continuum; otherwise we must use statistical
mechanics to obtain the governing equations of the problem whereby we can establish that:
Kn
Kn
/
1
macroscopic approach
"
/
!1
microscopic approach
"
The fundamental hypothesis in Continuum Mechanics is that the matter of which the
medium is made up is continuously distributed and that the variables involved in the
problem (e.g. velocity, acceleration, pressure, mass density, etc.) are continuous functions.
Then, by means of approximations or additional equations to that initially proposed for the
problem, we can characterize a continuum with discontinuous variables associated with the
problem, e.g. fracture problem and shock waves among others.
2.2
The Continuum
In general, when we apply force to solids they are able to recover their original states when
said force is removed. However, this is not the case with fluids, i.e. solids and fluids
apparently act very differently. Therefore, traditionally, continuum mechanics has been
divided into two groups: solids and fluids (liquids and gases). As we will see throughout this
book, the fundamental equations of Continuum Mechanics are the same for both of these.
For many decades, solid and fluid mechanics have been treated independently from each
other. However, nowadays, it is not advisable to work like this. Firstly, it is necessary to
simulate more complex materials, e.g. materials that have characteristics of solids and fluids
simultaneously. These materials, besides presenting elastic properties, (obeying the
constitutive law for solids), also exhibit characteristics of fluids due to their viscosity, for
example: viscoelastic materials. Secondly, the need to simulate the problem of fluid-solid
interaction has improved the relationship between fluids and solids.
Recently, a third branch of continuum mechanics has emerged, which is related to
multiphysics problems, characterized by phase change, e.g. from solid to liquid phase or
vice versa, and which includes mechanical systems that transcend classical mechanical
boundary of solids and fluids. Then, traditionally, continuum mechanics can be divided
into:
Solids
Liquids
Multiphysics
INTRODUCTION
Metric level
At this level, we include most problems posed in Civil, Mechanical, Aerospace
Engineering.
Millimeter level
At this level, it may enroll the specimen used to measure the material mechanical
properties in the laboratory.
Micrometer level
Micro-structural characteristic, such as micro-defects and cement hydration products,
are observed at this scale.
Nanometer level
At this level, we contemplate atomic and molecular processes.
1u 10 0 m
1 u 10 3 m
Structural Mechanics
Macro Mechanics
Meso Mechanics
1 u 10 6 m
Micro Mechanics
1 u 10 9 m
Nano Mechanics
3.1
The continuum mechanics is raised at a macroscopic level. That is, the variables of the
problem at a macroscopic level are considered as being the average of these variables at a
mesoscale level. Let us take, for example, blood, which can be treated in different ways,
depending on the scale under consideration. At a 10 6 m scale, we consider blood flows
around a blood cell where the deformation of the cell walls is taken into consideration.
Then, at a 10 4 m scale, we can consider the fluid flow through a set of blood cells, which
thus allows us to observe the fluid effects on cells. Next, at a 10 3 m scale (macroscopic
level), we can consider the fluid flow through arteries or veins (ignoring the individual cells)
as being a fluid with certain macroscopic properties (e.g. velocity, pressure, etc.), (see Figure
2).
INTRODUCTION
We can determine the mass density of a cube (with sides a ) by dividing its total mass by its
volume. So, let us consider a new cube (with sides ac ) whose volume is less than the first
one. In Figure 3(b) we can observe that, depending on the position of the new cube, we
can obtain different values for mass density, as different position contain different amounts
of matter and voids.
That is, if we can vary the a -dimension from a very small size, we will notice that the mass
density value will oscillate, (see Figure 4). However, there will be a a -dimension region in
which the mass density value maintains constant. The continuum mechanics is raised into
this interval.
...
...
...
...
ac
...
...
ac
(a)
(b)
Scale of
Continuum
Mechanics
log(a )
&
u
Analytical (exact)
IBVP
SOLUTION
Numerical
Figure 5: Statement and solution of the problem.
4.1
Once the physical problem is stated, it can be solved and the IBVP solution can be
analytical (exact solution), or numerical (approximated solution), (see Figure 5).
In practice, obtaining the analytical solution of the IBVP is very difficult or even
impossible because of the problem complexity (e.g. due to its geometry, forces, or
boundaries), hence we must resort to using IBVP numerical solution. However, obtaining
the analytical solution for simple problems is very important since it serves as a reference
to indicate the degree of accuracy (precision) of the numerical technique used.
Among the most widely used numerical techniques for the IBVP solution we can list,
among others:
INTRODUCTION
We cannot state that any one of the techniques mentioned above is the best. First we must
ask what type of problem we want to solve and, depending on this, one technique or
another, or even a combination of different techniques can be used to optimize the
solution.
In general, all techniques transform the continuous problem into a discrete system of
equations.
The finite difference method (FDM) is based on discretizing the domain by points in
which the governing equations are valid. The FDM was the first numerical method to
emerge and today it is still in use on problems where stabilization problems occur and is
also used to dicretize the time domain.
The finite element method (FEM) is based on discretizing the domain into subdomains
called finite elements, in which the governing equations are valid. Moreover, it has proved
to be more accurate in solving problems that FDM. Today the FEM technique is the most
used and widespread in the solid mechanics field.
Conversely, in the Boundary Element Method (BEM), only the domain boundary is
discretized by elements. From a viewpoint of the solution accuracy, the BEM provides
more accurate solutions than FEM for elastic problems and it is a better method for
working with semi-infinite or infinite problem domains. Nevertheless, the BEM has its
downside in nonlinear problems, where we need to dicretize the domain by cells.
Generally, the IBVP contains both spatial variables (displacement, pressure, etc.) and
temporary variables (rates of change of spatial variables), so we need to have both spatial
and time dicretization to obtain the numerical solutions. For example, for spatial
discretization we can use the FEM and for time discretization we can use another
technique, such as the FDM.
4.2
There are cases where the problem (IBVP) includes certain features which allow it to be
simplified whereby its complexity can be drastically reduced and with which even the
analytical problem solution can be obtained. These simplifications will be pointed out and
elaborated on in detail throughout this book, but the engineers will have to decide for
themselves when these simplifications can be used for a given problem and for this a sound
grounding in the general theory is needed.
1 Tensors
1
Tensors
1.1 Introduction
As seen previously in the introductory chapter, the goal of continuum mechanics is to
establish a set of equations that governs a physical problem from a macroscopic
perspective. The physical variables featuring in a problem are represented by tensor fields,
in other words, physical phenomena can be shown mathematically by means of tensors
whereas tensor fields indicate how tensor values vary in space and time. In these equations
one main condition for these physical quantities is they must be independent of the
reference system, i.e. they must be the same for different observers. However, for matters
of convenience, when solving problems, we need to express the tensor in a given
coordinate system, hence we have the concept of tensor components, but while tensors are
independent of the coordinate system, their components are not and change as the system
change.
In this chapter we will learn the language of TENSORS to help us interpret physical phenomena.
These tensors can be classified according to the following order:
Zeroth-Order Tensors (Scalars): Among some of the quantities that have magnitude but
not direction are e.g.: mass density, temperature, and pressure.
First-Order Tensors (Vectors): Quantities that have both magnitude and direction, e.g.:
velocity, force. The first-order tensor is symbolized with a boldface letter and by an arrow
&
at the top part of the vector, i.e.: x .
Second-Order Tensors: Quantities that have magnitude and two directions, e.g. stress and
strain. The second-order and higher-order tensors are symbolized with a boldface letter.
In the first part of this chapter we will study several tools to manage tensors (scalars,
vectors, second-order tensors, and higher-order tensors) without heeding their dependence
E.W.V. Chaves, Notes on Continuum Mechanics, Lecture Notes on Numerical
Methods in Engineering and Sciences 4, DOI 10.1007/978-94-007-5986-2_2,
International Center for Numerical Methods in Engineering (CIMNE), 2013
10
on space and time. At the end of the chapter we will introduce tensor fields and some field
operators which can be used to interpret these fields.
In this textbook we will work indiscriminately with the following notations: tensorial,
indicial, and matricial. Additionally, when the tensors are symmetrical, it is also possible to
represent their components using the Voigt notation.
&
Addition: Let a , b be arbitrary vectors, we can show the sum of adding them, (see Figure
&
1.1 (a)), with a new vector ( c ) thus defined as:
& & & & &
c ab ba
&
a
(1.1)
&
d
&
c
&
a
&
b
&
b
a)
&
c
&
b
b)
Figure 1.1: Addition and subtraction of vectors.
& &
Subtraction: The subtraction between two arbitrary vectors ( a , b ), (see Figure 1.1 (b)), is given
as follows:
(1.2)
(1.3)
&
Scalar multiplication: Let a be a vector, we can define the scalar multiplication with Oa .
&
The product of this operation is another vector with the same direction of a , and whose
length and orientation is defined with the scalar O as shown in Figure 1.2.
O 1
&
a
O !1
&
a
O0
&
Oa
0 O 1
&
a
&
Oa
&
a
&
Oa
1 TENSORS
11
Scalar Product: The Scalar Product (also known as the dot product or inner product) of two
& &
& &
vectors a , b , denoted by a b , is defined as follows:
& &
& &
a b cos T
H ab
(1.4)
where T is the angle between the two vectors, (see Figure 1.3(a)), and x represents the
Euclidean norm (or magnitude) of x . The result of the operation (1.4) is a scalar.
& &
& & & &
Moreover, we can conclude that a b b a . The expression (1.4) is also true when a b ,
therefore:
& &
aa
& &
& &
0
a a cos T T
o aa
&
& &
&
a a a
& &
aa
(1.5)
& &
aa .
&
Unit Vector: A unit vector, associated with the a -direction, is shown with a a , which has
&
the same direction and orientation of a . In this textbook, the hat symbol ( x ) denotes a
&
unit vector. Thus, the unit vector, a , codirectional with a , is defined as:
&
a
&
a
(1.6)
&
&
where a represents the norm (magnitude) of a . If a is the unit vector, then the
following must be true:
a
(1.7)
(1.8)
&
&
Projection Vector: The projection vector of a onto b , (see Figure 1.3(b)), is defined as:
&
proj b& a
&
&
&
proj b& a b Projection vector of a onto b
(1.9)
&
&
&
where proj b& a is the projection of a onto b , and b is the unit vector associated with the
&
&
b -direction. The magnitude of proj b& a is obtained by means of the scalar product:
&
proj b& a
&
&
&
a b Projection of a onto b
(1.10)
So, taking into account the definition of the unit vector, we obtain:
&
proj b& a
& &
ab
&
b
(1.11)
&
& &
ab
& b
b
& &
ab
&
b
&
b
&
b
& &
ab &
& 2 b
b
,
scalar
(1.12)
12
0dTdS
&
a
T
& &
ab
&
a
T
&
b
&
projb& a
&
b
& &
a b cos T
a) Scalar product
b) Projection vector
&
Orthogonality between vectors: Two vectors a and b are orthogonal if the scalar product
between them is zero, i.e.:
& &
ab 0
(1.13)
& &
Vector Product (or Cross Product): The vector product of two vectors, a , b , results in
&
another vector c , which is perpendicular to the plane defined by the two input vectors,
(see Figure 1.4). The vector product has the following characteristics:
Representation:
& &
& & &
c a b b a
(1.14)
&
&
&
The vector c is orthogonal to the vectors a and b , thus:
& & & &
ac b c 0
&
The magnitude of c is defined by the formula:
&
& &
c
a b sin T
(1.15)
(1.16)
&
&
where T measures the smallest angle between a and b , (see Figure 1.4).
&
&
The magnitude of the vector product a b is geometrically expressed as the area of the
parallelogram defined by the two vectors, (see Figure 1.4):
A
& &
ab
(1.17)
Therefore, the triangle area defined by the points OCD , (see Figure 1.4 (a)), is:
1 & &
ab
(1.18)
2
&
&
&
&
If a and b are linearly dependent, i.e. a Bb with B denoting a scalar, the vector product
& & &
& &
of two linearly dependent vectors becomes a zero vector, a b Bb b 0 .
& & &
Scalar Triple Product (or Mixed Product): Let a , b , c be arbitrary vectors, we can
AT
(1.19)
1 TENSORS
13
where the scalar V represents the volume of the parallelepiped defined by a, b, c , (see
Figure 1.5).
&
c
& &
ab
&
b
D
A
..
&
b
&
c
&
a
a)
&
c
AT
..
&
a
& &
ba
b)
If two vectors are linearly dependent then, the scalar triple product is zero, i.e.:
&
0
(1.20)
&
c
&
b
T
..
&
a
&
&
Vector Triple Product: Let a , b , c be vectors, we can define the vector triple product as
& & & &
w a b c . Then, we can demonstrate that the following relationships to be true:
&
w
(1.22)
&
whereby it is clear that the result of the vector triple product is another vector w , belonging to
&
&
the plane 31 formed by the vectors b and c , (see Figure 1.6).
14
& &
31 - plane defined by b , c
31
32
&
c
&
b
&
w belonging to the plane 31
&
a
& &
bc
&
w
&
Problem 1.1: Let a and b be arbitrary vectors. Prove that the following relationship is
true:
Solution:
a ab b a b
& & & &
& &
& & 2
ab
2
& &
a b sin T
& 2 & 2
a b sin 2 T
& 2 & 2
a b 1 cos 2 T
& 2 & 2
& 2 & 2
a b a b cos 2 T
2
& 2 & 2
& &
a b a b cos T
& 2 & 2 & & 2
a b ab
& & & &
& & 2
a a b b a b
Linear Transformation
&
&
& &
&
&
F (u v ) F (u) F ( v )
&
&
F (Bu) BF (u)
1 TENSORS
15
1 2
EH , demonstrate
2
H1
H2
H1 H 2
1 2
EH does not show a linear transformation because the condition
2
Z (H1 H 2 ) Z (H1 ) Z (H 2 ) has not been satisfied:
1
1
1 2 1 2 1
2
E >H 1 H 2 @
E H 12 2H1 H 2 H 22
EH 1 E H 2 E 2 H 1 H 2
Z (H1 H 2 )
2
2
2
2
2
Z ( H 1 ) Z ( H 2 ) EH 1 H 2 z Z ( H 1 ) Z ( H 2 )
>
Z ( H)
Z (H1 H 2 )
Z (H1 ) Z (H 2 )
Z (H 2 )
Z (H1 )
H1
H2
H1 H 2
16
COMPONENTS
Tensor Representation in a
Coordinate System
Figure 1.7: Tensor components.
&
Let a be a first-order tensor (vector) as shown in Figure 1.8 (a), the tensor representation
in a general coordinate system, defined as [1 , [ 2 , [ 3 , is made up of its components
( a1 , a 2 , a 3 ), (see Figure 1.8 (b)). Some examples of coordinate system are: the Cartesian
coordinate system; the cylindrical coordinate system; and the spherical coordinate system.
[3
[2
&
a ( a1 , a 2 , a 3 )
&
a
&
a
b)
a)
[1
1.3.1
The Cartesian coordinate system is defined by three unit vectors: i , j , k , denoted by the
Cartesian basis, which make up an orthonormal basis. The orthonormal basis has the
following properties:
1. The vectors that make up this basis are unit vectors:
i
(1.23)
or:
i i j j k k 1
(1.24)
1 TENSORS
17
k i 0
(1.25)
j k
k i j
(1.26)
The direction and orientation of the orthonormal basis can be obtained using the righthand rule as shown in Figure 1.9.
i j k
j k
k i j
j
j
i
i
k
(1.27)
1.3.2
The vector a , (see Figure 1.10), in the Cartesian coordinate system, is represented by its
different components ( a x , a y , a z ) and by the Cartesian bases ( i , j , k ) as:
&
a a x i a y j a z k
(1.28)
y
ay
&
j a
k
ax
x
az
z
&
&
Let a , b , c be arbitrary vectors, we can describe some vector operations in the Cartesian
coordinate system, as follows:
18
& &
(1.29)
& 2
a .
a 2x a 2y a 2z
NOTE: The projection of a vector onto a given direction was established in the
equation (1.10), thus defining the component concept. For example, if we want to
know the vector component along the y -direction, all we need to do is calculate:
&
a j (a x i a y j a z k ) (j) a y .
&
a 2x a 2y a 2z
a 2y a 2z
&
0 0 i 0 j 0 k
&
&
Addition: The vector sum of a and b is represented by:
(1.32)
a x b x i a y b y j a z b z k
a x b x i a y b y j a z b z k
(1.33)
(1.34)
&
(1.31)
&
&
Subtraction: The difference between a and b is:
& &
a b (a x i a y j a z k ) (b x i b y j b z k )
az
a 2x
& &
a b (a x i a y j a z k ) (b x i b y j b z k )
(1.30)
&
Then, the unit vector codirectional with a is:
&
ay
ax
a
i
j
a
&
2
2
2
2
a
ax a y az
a x a 2y a 2z
Oa x i Oa y j Oa z k
& &
The vector product ( a b ) is evaluated as:
&
c
& &
ab
i
ax
j
ay
k
az
bx
by
bz
ay
by
az
i a x
bz
bx
(1.35)
az ax
j
bx
bz
ay
k
by
(1.36)
(a y b z a z b y )i (a xb z a z b x )j (a xb y a y b x )k
The scalar triple product [a, b, c] is the determinant of the 3 by 3 matrix, defined
as:
1 TENSORS
by
bz
cy
cz
19
ay
cx
bz
az
cz
bx
by
cx
cy
ax
ay
az
bx
cx
by
cy
bz
cz
(1.37)
a x b y c z b z c y a y b x c z b z c x a z b x c y b y c x
& & &
The vector triple product made up of the vectors ( a, b, c ) is obtained, in the
a c b a bc
& & &
(1.38)
O 1b x O 2 c x i O 1b y O 2 c y j O 1b z O 2 c z k
& &
where O 1 a c a x c x a y c y a z c z , and O 2
& &
a b a xb x a yb y a zb z .
Problem 1.3: Consider the points: A1,3,1 , B 2,1,1 , C 0,1,3 and D1,2,4 , defined in the
Cartesian coordinate system.
o
1) Find the parallelogram area defined by AB and AC ; 2) Find the volume of the
o
BC .
Solution:
o
2i 1j 1k 1i 3j 1k
0i 1j 3k 1i 3j 1k
1i 4j 0k
1i 2j 2k
With reference to the equation (1.36) we can evaluate the vector product as follows:
i
1
& &
ab
j k
4 0
1 2
( 8)i 2j ( 6)k
Then, the parallelogram area can be obtained using definition (1.19), thus:
& &
ab
104
AD OD OA
1i 2j 4k 1i 3j 1k
0i 1j 3k
and using the equation (1.37) we can obtain the volume of the parallelepiped:
& & &
V (a, b, c )
0i 1j 3k 8i 2j 6k
0 2 18
16
BC
OC OB
0i 1j 3k 2i 1j 1k
2i 2j 2k
20
proj BCo AB
BC AB
BC
BC
BC
1.3.3
2i 2j 2k 1i 4j 0k 2i 2j 2k
2i 2j 2k 2i 2j 2k
BC
2 8 0 2i 2j 2k
4 4 4
5 5 5
i j k
3
3
3
As we saw in equation (1.28) a in the Cartesian coordinate system was defined as:
&
a a x i a y j a z k
(1.39)
&
a a1e 1 a 2 e 2 a 3 e 3
(1.40)
a e
i
(1.41)
i 1
Then, we introduce the summation convention, according to which the repeated indices
indicate summation. So, equation (1.41) can be represented as follows:
&
a a1e 1 a 2 e 2 a 3 e 3 a i e i
(i 1,2,3)
&
a a i e i
(i 1,2,3)
(1.42)
NOTE: The summation notation was introduced by Albert Einstein in 1916, which led to
the indicial notation.
(1.43)
1 TENSORS
21
ai
a1
a
2
a 3
(1.44)
y { x2
a y { a2
&
a
j { e
2
a x { a1
i { e
1
x { x1
k { e 3
a z { a3
z { x3
Unit vector components: Let a be a vector, the normalized vector a is defined as:
a
&
a
&
a
with
(1.45)
ai
a12
a 22
a 32
ai
ai
a ja j
ak ak
(i, j , k 1,2,3)
(1.46)
22
Scalar product: Using definitions (1.4) and (1.29), we can express the scalar product
& &
( a b ) as follows:
& &
H ab
& &
a b cos T a1b1 a 2 b 2 a 3 b 3
aibi
a jb j
(1.47)
(i, j 1,2,3)
bx
a 31 x a 32 y a 33 z
bz
3) a 21 x a 22 y a 23 z b y
Solution:
a11 x1 a12 x 2 a13 x 3
a 21 x1 a 22 x 2 a 23 x 3
a x a x a x
32 2
33 3
31 1
b1
b2
b3
a1 j x j
o a 2 j x j
a3 j x j
dummy
index j
b1
b2
free
index
i o
a ij x j
bi
b3
As we can appreciate in this problem, the use of the indicial notation means that the
equation becomes very concise. In many cases, if algebraic operation do not use indicial or
tensorial notation they become almost impossible to deal with due to the large number of
terms involved.
Problem 1.5: Expand the equation: Aij x i x j
(i, j 1,2,3)
Solution: The indices i, j are dummy indices, and indicate index summation and there is no
free index in the expression Aij x i x j , therefore the result is a scalar. So, we expand first the
dummy index i and later the index j to obtain:
expanding j
i o A1 j x1 x j A2 j x 2 x j A3 j x 3 x j
Aij x i x j expanding
A11 x1 x1 A21 x 2 x1 A31 x 3 x1
A12 x1 x 2
A22 x 2 x 2
A32 x 3 x 2
A13 x1 x 3
A23 x 2 x 3
A33 x 3 x 3
1.4.1
Some Operators
1.4.1.1
Kronecker Delta
1 TENSORS
1 iff
0 iff
E ij
23
(1.48)
iz j
Also note that the scalar product of the orthonormal basis e i e j is equal to 1 if i
j and
e i e j
e 1 e 2
e 2 e 2
e 3 e 2
e 1 e 3
e 2 e 3
e 3 e 3
1 0 0
0 1 0 E
ij
0 0 1
(1.49)
An interesting property of the Kronecker delta is shown in the following example. Let Vi
&
E 1 jV1 E 2 jV 2 E 3 jV3
(1.50)
2 E ij Vi
3 E ijVi
V3
(1.51)
That is, in the presence of the Kronecker delta symbol we replace the repeated index as
follows:
Ei
Vj
(1.52)
For this reason, the Kronecker delta is often called the substitution operator.
Other examples using the Kronecker delta are presented below:
E ij Aik
A jk , E ij E ji
E ii
E jj
E 11 E 22 E 33 3 , E ji a ji
aii a11 a 22 a 33
(1.53)
&
To obtain the components of the vector a in the coordinate system represented by e i , it
&
&
is sufficient to obtain the scalar product with a and e i , i.e. a e i a p e p e i a p E pi a i .
With that, it is also possible to represent the vector as:
&
a a i e i
&
(a e i )e i
(1.54)
3 z E 11 1 ,
since what must be replaced is the repeated index, not the number
1.4.1.2
Permutation Symbol
The permutation symbol . ijk (also known as Levi-Civita symbol or alternating symbol) is defined as:
24
. ijk
(1.55)
k ) or (i
k)
NOTE: . ijk are the components of the Levi-Civita pseudo-tensor, which will be introduced
later on.
The values of . ijk can be easily memorized using the mnemonic device shown in Figure
1.12(a), in which if the index values are arranged in a clockwise direction, the value of . ijk
is equal to 1 , if not it has the value of 1 . In the same way we can use this mnemonic
device to switch indices, (see Figure 1.12(b)).
.ijk
.ijk
i
1
.ijk
.ikj
.ijk
. jki
. ijk
. ikj
.kij
. kji
. jik
b)
a)
. ijk
1
(i j )( j k )(k i )
2
(1.56)
Using both the definition seen in (1.55) and Figure 1.12 (b), it is possible to verify that the
following relations are valid:
. ijk
. jki
. ijk
. ikj
. kij
. jik
. kji
(1.57)
. lmn E li E mj E nk
E 1i E 2 j E 3k E 1i E 3 j E 2k E 2i E 1 j E 3k E 3i E 1 j E 2 k E 2i E 3 j E 1k E 3i E 2 j E 1k
E 1i E 2 j E 3k E 3 j E 2 k E 1 j E 2i E 3k E 3i E 2 k E 1k E 2i E 3 j E 3i E 2 j
(1.58)
. ijk
E 1i E 1 j E 1k
E 2i E 2 j E 2 k
E 3i E 3 j E 3k
E 1i E 2i E 3i
E1 j E 2 j E 3 j
E 1k E 2 k E 3k
(1.59)
. ijk . pqr
E 1i
E1 j
E 1k
E 2i E 3i E 1 p E 1q E 1r
E 2 j E 3 j E 2 p E 2q E 2r
E 2 k E 3k E 3 p E 3q E 3r
(1.60)
1 TENSORS
25
Taking into account that det (AB ) det (A )det (B ) , where det (x) { x is the determinant
of the matrix x , the equation (1.60) can be rewritten as:
. ijk . pqr
E 1i
E
1j
E 1k
E 3i E 1 p E 1q E 1r
E 3 j E 2 p E 2 q E 2 r
E 3k E 3 p E 3q E 3r
E 2i
E2j
E 2k
. ijk . pqr
E ip E iq E ir
E jp E jq E jr
E kp E kq E kr
(1.61)
E mi E mp
For the special exception when r k , the equation (1.61) is reduced to:
E ip E iq E ik
E jp E jq E jk
E kp E kq 3
. ijk . pqr
. ijk . pqk
E ip E jq E iq E jp
i, j , k , p, q 1,2,3
(1.62)
6 . b) Obtain the
.123 1
&
&
&
& &
ab
e 1
e 2
e 3
a1 a 2 a 3
b1 b 2 b 3
(a 2 b 3 a 3b 2 )e 1 (a 3b1 a1b 3 )e 2 (a1b 2 a 2 b1 )e 3
c1
(1.63)
c3
c2
Using the definition of the permutation symbol . ijk , defined in (1.55), we can express the
&
components of c as follows:
.1 jk a j b k
c1
.123 a 2 b 3 .132 a 3b 2
c2
c3
. 3 jk a j b k
. ijk a j b k
(1.64)
&
Then, the vector product ( a b ) can be represented by means of the permutation symbol
as:
& &
a b . ijk a j b k e i
a j e j b k e k
a j b k . ijk e i
a j b k (e j e k ) a j b k . ijk e i
(1.65)
a j b k . jki e i
26
(1.66)
The permutation symbol and the orthonormal basis can be interrelated using the triple
scalar product as follows:
e i e j
. ijm e m
e j
. ijm E mk e i e j e k
. ijm e m e k
. ijk
(1.67)
(1.68)
(1.69)
or
a1 a 2
b1 b 2
a3
b3
c1
c3
c2
(1.70)
Starting from the equation (1.69) we can prove the following are true:
. jki aib j c k
. kij aib j c k
.ikj aib j c k
. jik aib j c k
. kji aib j c k
(1.71)
where we take into account the property of the permutation symbol as given in (1.57).
Problem 1.8: Rewrite the expression a b c d without using the vector product
symbol.
& &
Solution: The vector product a b can be expressed as
& &
ab
& &
cd
a j e j b k e k
& &
. ijk a j b k e i . Likewise, it is possible to express c d as
. nlm c l d m e n , thus:
Taking into account that . ijk . ilm . jki . lmi (see equation (1.57)) and by applying the
equation (1.62), i.e.: . jki . lmi E jl E km E jm E kl . jki . ilm , we obtain:
. ijk . ilm a j b k c l d m E jl E km E jm E kl a j b k c l d m a l b m c l d m a m b l c l d m
Since a l c l
& &
& &
b d holds true, we can conclude that:
a& b& c& d& a&& c& &b& d& a& d& b& c&
a c and b m d m
&
&
1 TENSORS
& &
ab
27
a a b b a b b a
& & & &
&
a
&
b
& &
ab
&
&
@ & >&
&
&
>
@ >
p
. ijk a j b k c p d i . ijk a j b k c i d p
. ijk a j b k c p d i c i d p
>
>
&
and
c& d& ,
&
a& b&
&
&
&
&
v1
b1
v2
b2
v3
b3
v1
v2
b1
b2
c1
c2
c1
c2
c3
v3
b3
c3
a1
b1
a2
b2
a3
b3
a1
a2
b1
b2
c1
c2
c1
c2
c3
a3
b3
c3
. ijk v i b j c k
. pqr a p b q c r
& &
&
Solution: Taking into account that (d) b c
i
obtain:
& &
. qjk b j c k , we
28
& &
& &
a k b q c k a j b j c q b q a c c q a b
& & & && &
& & &
a b c q ba c c a b q
>
@ >
Dyadic
&
&
The tensor product, made up of two vectors v and u , becomes a dyad, which is a particular
case of a second-order tensor. The dyad is represented by:
&& & &
uv { u v
(1.72)
where the operator denotes the tensor product. Then, we define a dyadic as a linear
combination of dyads. Furthermore, as we will see later, any tensor can be represented by
means of a linear combination of dyads, (see Holzapfel (2000)).
The tensor product has the following properties:
1.
2.
3.
(1.73)
(1.74)
(1.75)
where B and C are scalars. By definition, the dyad does not contain the commutative
& & & &
property, i.e., u v z v u .
The equation (1.72) can also be expressed in the Cartesian system as:
& &
A uv
A
,
Tensor
(u i e i ) ( v j e j )
u i v j (e i e j )
A ij (e i e j )
(i, j 1,2,3)
(1.76)
A ij e i e j
,
(i, j 1,2,3)
(1.77)
components
basis
u
v
p
components
p
( A ) ij
& &
(u v ) ij
(1.78)
ui v j
A ij
1 TENSORS
( A ) ij
A ij
A 11
A 12
A 31
A 32
29
A 13
A 23
A 33
A A 21 A 22
(1.79)
It is easy to identify the tensor order by the number of free indices in the tensor
components, i.e.:
Second-order tensor U U ij e i e j
Third-order tensor
Tijk e i e j e k
Fourth-order tensor
I ijkl e i e j e k e l
(i, j , k , l 1,2,3)
(1.80)
OBS.: The tensor order is given by the number of free indices in its components.
OBS.: The number of tensor components is given by a n , where the base a is the
maximum value in the index range, and the exponent n is the number of the free
index.
Problem 1.12: Define the order of the tensors represented by their Cartesian components:
v i , ) ijk , Fijj , H ij , C ijkl , V ij . Determine the number of components in tensor C .
Solution: The order of the tensor is given by the number of free indices, so it follows that:
& &
First-order tensor (vector): v , F ; Second-order tensor: , ; Third-order tensor: ) ;
Fourth-order tensor: C
The number of tensor components is given by the maximum index range value, i.e.
i, j , k , l 1,2,3 , to the power of the number of free indices which is equal to 4 in the case of
C ijkl . Thus, the number of independent components in C is given by:
34
3 u j
3 u k
3 u l
3 81
Addition: The sum of two tensors of same order is a new tensor defined as follows:
C
A B B A
(1.81)
( A B) ij
or
C ij
A ij B ij
(1.82)
C AB
(1.83)
O( A ) ij
(1.84)
30
A 11
A 12
A 31
A 32
OA 11
O A
21
OA 31
A 13
A 23
o OA
A 33
A A 21 A 22
OA 12
OA 22
OA 32
OA 13
OA 23
OA 33
(1.85)
&
O( A v )
(1.86)
&
for any vector v .
Scalar Product (or Dot Product): The scalar product (also known as single
&
contraction) between a second-order tensor A and a vector x is another vector
&
(first-order tensor) y , defined as:
&
y
E kl
&
Ax
( A jk e j e k ) ( x l e l )
A jk x l E kl e j
A jk x k e j
(1.87)
yj
y j e j
The scalar product between two second-order tensors A and B is another second-order
tensor, that verifies: A B z B A :
E jk
C
A B
E jk
( A ij e i e j ) (B kl e k e l )
A ij B kl E jk e i e l
A B kl e i e l
ik
D BA
AB
(B ij e i e j ) ( A kl e k e l )
B ij A kl E jk e i e l
B ik A kl e i e l
(1.88)
BA
C il e i e l
D il e i e l
A B A C
A (B C) ( A B) C
(1.89)
The scalar product allows us to define the power of second-order tensors, as seen below:
A0
1 ; A1
A ; A2
AA
A3
A A A , and so on,
(1.90)
where 1 is the second-order unit tensor (also called the identity tensor).
&
&
&
In components
c v& d u&
&
&
(1.91)
1 TENSORS
31
E il
E jk
A B
( A ij e i e j ) (B kl e k e l )
A ij B kl E jk E il
(1.92)
A ij B ji
H ( scalar )
A :B
c u& d v&
&
&
(1.93)
u v : c d u c v d c ud v
&
&
&
&
A :B
(1.94)
( A ij e i e j ) : (B kl e k e l )
E jl
A ij B kl E ik E jl
A ij B ij
O
(1.95)
( scalar )
In general, A : B z A B , however, they are equal if at least one of them is symmetric, i.e.
A sym : B A sym B or A : B sym A B sym , so A sym : B sym A sym B sym .
The double contraction with a third-order tensor ( S ) and a second-order tensor ( B )
becomes:
S :B
B :S
a v& d u& c
& &
&
u& c v& da
&
&
&
(1.96)
As we can verify the result is a vector. In symbolic notation, the double contraction ( B : S )
is represented by:
S ijk e i e j e k : B pq e p e q
S ijk B pq E jp E kq e i
S ijk B jk e i
(1.97)
C ijkl H pq E kp E lq e i e j
C ijkl H kl e i e j
V ij e i e j
(1.98)
32
(1.99)
demonstrate that:
&
&
a A b a p e p A ij e i e j b r e r
& &
A : ( a b)
a p A ij b r E pi E jr
a i A ij b j
A ij (a i b j )
(1.101)
Vector product
&
( A ij e i e j ) ( x k e k ) . ljk A ij x k e i e l
(1.102)
>
>
>(a
k a k )E jp ( a k E kp ) a j b p
&
& &
>(a a)1 a& a& @ b j
>(a
k ak
)E jp a p a j b p
(1.103)
(1.104)
1.5.1.1
a c b a b c b c c b a
&
>(a& a& )1 a& a& @ b
& & &
&
&
&
&
&
(1.105)
1 TENSORS
Tij e i e j
33
Ti1e i e 1 Ti 2 e i e 2 Ti 3 e i e 3
(1.106)
Tij e i e j e k
Tij e i E jk
Tik e i
T1k e 1 T2 k e 2 T3k e 3
(1.107)
& ( e )
T e k Tik e i
(1.108)
k 2 Ti 2 e i T12 e 1 T22 e 2 T32 e 3 t 2
& (e )
3
k 3 Ti 3 e i T13 e 1 T23 e 2 T33 e 3 t
& &
&
Graphical representation of these three vectors t ( e1 ) , t (e 2 ) , t (e 3 ) , in the Cartesian basis, is
& ( e )
shown in Figure 1.13. Note also that t 1 is the projection of T onto e 1 , n (i1) >1,0,0@ ,
T11
T
21
T31
T12
T22
T32
T13 1
T23 0
T33 0
T11
T
21
T31
x3
&
t (e3 )
e 3
&
t ( e1 )
t i( e1 )
(1.109)
&
t (e 2 )
e 2
x2
e 1
x1
>
(1.110)
34
components. The components displayed tangentially to the plane are called tangential
components, and correspond to the off-diagonal terms of Tij .
x3
Tij
T11
T
21
T31
T12
T22
T32
&
t (e 3 )
T33 e 3
T13
T23
T33
&
t (e 2 )
T23 e 2
T13 e 1
T32 e 3
T31e 3
&
t ( e1 )
T21e 2
T22 e 2
T12 e 1
x2
T11e 1
x1
Symbolic notation
A B
( A ij e i e j ) (B kl e k e l )
(1.111)
A ij B kl E jk (e i e l )
A ij B jl (e i e l )
Cartesian basis
Indicial notation
Note that the index is not repeated more than twice either in symbolic notation or in
indicial notation. Also note that the indicial notation is equivalent to the tensor notation
& &
only when dealing with scalars, e.g. A : B A ijB ij O , or a b a i b i .
1.5.2
1.5.2.1
Properties of Tensors
Tensor Transpose
A ji (e i e j )
A ij (e j e i )
(1.112)
If A ij are the components of A , it follows that the components of the transpose are:
1 TENSORS
(A T ) ij
35
A ji
(1.113)
& &
& &
If A u v , the transpose of the dyad A is given by A T v u :
& &
& &
u v T
AT
v u
T
u i e i v j e j
v j e j u i e i
v i e i u j e j
T
u i v j e i e j
u i v j e j e i
u j v i e i e j
T
A ij e i e j
A ij e j e i
A ji e i e j
(1.114)
; (BB CA ) T
A
A
AT :B
e : B
BB T CA T
; (B A ) T
A T BT
e j : B kl e l e k A ij B kl E il E jk
A ij B ji
A B
A ij B ji
A B
ij e i
ij e j
e l A ij B kl E jk E il
kl e k
(1.115)
(1.116)
The transpose of the matrix A is formed by changing rows for columns and vice versa,
i.e.:
A
A 11
A
21
A 31
A 12
A 22
A 32
A 13
o A T
A 23 transpose
A 33
A 11
A
21
A 31
A 12
A 22
A 13
A 23
A 33
A 32
A 11
A
12
A 13
A 21
A 22
A 23
A 31
A 32
A 33
(1.117)
A : B C B T A : C A C T : B
Solution: Expressing the term A : B C in indicial notation we obtain:
A : B C
A ij e i e j : B lk e l e k C pq e p e q
A ij B lk C pq e i e j : E kp e l e q
A ij B lk C pq E kp E il E jq A ij B ik C kj
Note that, when we are dealing with indicial notation the position of the terms does not
matter, i.e.:
A ij B ik C kj
B ik A ij C kj
A ij C kj B ik
We can now observe that the algebraic operation B ik A ij is equivalent to the components of
the second-order tensor (B T A ) kj , thus,
(B T A ) kj C kj
B ik A ij C kj
A : C .
A C :B .
& &
Problem 1.14: Let u , v be vectors and A be a second-order tensor. Show that the
&
&
u AT v
&
&
v A u
&
&
u AT v
u i e i A jl e l e j v k e k
&
&
v A u
v k e k A jl e j e l u i e i
u i A jl E il v k E jk
u l A jl v j
v k E kj A jl u i E il
v j A jl u l
36
1.5.2.2
1.5.2.2.1
Symmetric tensor
A second-order tensor A is symmetric, i.e.: A { A sym , if the tensor is equal to its transpose:
A
if
A T incomponents
o A ij
A ji
A is symmetric
(1.118)
A 11
A
12
A 13
A 12
A 22
A 13
A 23
A 33
(1.119)
in matrix form:
o A { A sym
A AT
A 23
From the above it is clear that a symmetric second-order tensor has 6 independent
components, namely: A 11 , A 22 , A 33 , A 12 , A 23 , A 13 .
According to equation (1.118), a symmetric tensor can be represented by:
A ij
A ji
A ij A ij
2 A ij
A ij
A ij A ji
(1.120)
A ij A ji
1
( A ij A ji )
2
1
(A A T )
2
A fourth-order tensor C , whose components are C ijkl , may have the following types of
symmetries:
Minor symmetry:
C ijkl
C jikl
C ijlk
C ijkl
C klij
(1.121)
C jilk
Major symmetry:
(1.122)
A fourth-order tensor that does not exhibit any kind of symmetry has 81 independent
components. If the tensor C has only minor symmetry, i.e. symmetry in ij ji (6) , and
symmetry in kl lk (6) , the tensor features 36 independent components. If besides
presenting minor symmetry it also provides major symmetry, the tensor features 21
independent components.
1.5.2.2.2
Antisymmetric tensor
A tensor A is antisymmetric (also called skew-symmetric tensor or skew tensor), i.e.: A { A skew :
if
A T incomponents
o A ij
A ji
A is antisymmetric
(1.123)
o A
A A
skew
0
A
12
A 13
A 12
0
A 23
A 13
A 23
0
(1.124)
1 TENSORS
37
Under the conditions expressed in (1.123), an antisymmetric tensor can be represented by:
A ij A ij
A ij A ji
A ij A ji
2 A ij
(1.125)
1
( A ij A ji )
2
A ij
1
(A A T )
2
Wij
1
(Wkl E ik E jl Wkl E jk E il )
2
1
Wkl (E ik E jl E jk E il )
2
(1.126)
Using the relation between the Kronecker delta and the permutation symbol given by
(1.62), i.e. E ik E jl E jk E il . ijr . lkr , the equation (1.126) is rewritten as:
Wij
1
Wkl . ijr . lkr
2
(1.127)
Expanding the term Wkl . lkr , for the dummy indices ( k , l ), we can obtain the following
nonzero terms:
W12 . 21r W13 . 31r W21.12 r W23 . 32 r W31.13r W32 . 23r
Wkl . lkr
(1.128)
thus,
r 1
Wkl . lkr
W23 W32
Wkl . lkr
W13 W31
Wkl . lkr
W12 W21
2 w1
2W12 2w3
2W23
2wr
(1.129)
0
0 w2
W32
w1
W13 W23
&
Hence, we introduce the axial vector w associated with the antisymmetric tensor, W , as:
&
w w1e 1 w2 e 2 w3 e 3
(1.131)
&
The magnitude of the axial vector w is given by:
& 2 & &
2
(1.132)
X2 w
w w w12 w22 w32 W23
W132 W122
Wij
0
W
21
W31
W12
0
W13
W23
0
. rij we obtain:
wr . rij
(1.133)
. kij Wij
wr . rij . kij
2 wr E rk
2 wk
(1.134)
38
1
. kij Wij
2
wk
(1.135)
Wij
0
W
12
W13
W12
0
W13
W23
0
W23
&
w
W12
w1e 1 w2 e 2 w3 e 3
W13
W23
W12
W12
w1
W23
x2
W23
w2
W13
W13
x1
&
(1.136)
&
&
& &
a W a W : (a a) 0
(1.137)
& &
when a b , it holds that:
& &
NOTE: Note that (a a) is a symmetric second-order tensor. Later on we will show that
the result of the double contraction between a symmetric tensor and an antisymmetric
tensor equals zero.
&
Wi1a1 Wi 2 a 2 Wi 3 a 3
(1.138)
Bearing in mind that the normal components are equal to zero for an antisymmetric tensor,
i.e., W11 0 , W22 0 , W33 0 , the scalar product (1.138) becomes:
i 1 W12 a 2 W13 a 3
&
W ai i 2 W21a1 W23 a 3
i 3 W a W a
31 1
32 2
(1.139)
&
&
The above components are the same as the result of the algebraic operation w a :
1 TENSORS
& &
wa
e 1
e 2
e 3
w1
a1
w2
a2
w3
a3
39
w3 a 2 w2 a 3 e 1 w3 a1 w1a 3 e 2 w2 a1 w1a 2 e 3
W12 a 2 W13 a 3 e 1 W21a1 W23 a 3 e 2 W31a1 W32 a 2 e 3
(1.140)
where w1 W23 W32 , w2 W13 W31 , w3 W12 W21 . Then, given an antisymmetric
&
tensor W and the axial vector w associated with W , it holds that:
& & &
W a w a
(1.141)
&
for any vector a . The property (1.141) could easily have been obtained by taking into
account the components of W given by (1.133), i.e.:
&
W ai
w j . jik a k
Wik a k
&
w& ai
. ijk w j a k
&
&
(1.142)
e 1*
(1.143)
Additionally, we can choose two unit vectors e *2 , e *3 , which make up an orthonormal basis
with the unit vector e 1* , (see Figure 1.16), so that:
e 1*
e *2 e *3
e *2
e *2
e *3 e 1*
e *3
&
w
e 3
e 2
e 1* e *2
(1.144)
Xe 1*
e 1*
e 1
e *3
&
By representing the vector a in this new basis, a a1* e 1* a*2 e *2 a*3 e *3 , the relationship
shown in (1.143) obtains the form below:
&
&
W a Xe 1* a Xe 1* (a1* e 1* a *2 e *2 a *3 e *3 )
>
e *3
&
X (e *3 e *2 e *2 e *3 ) a
e *2
(1.145)
Thus, an antisymmetric tensor can be represented, in the space defined by the axial vector,
as follows:
W
X (e *3 e *2 e *2 e *3 )
(1.146)
Note that by using the antisymmetric tensor representation shown in (1.146), the
projections of the tensor W according to directions e 1* , e *2 and e *3 are respectively:
40
W e 1*
&
0
W e *2
Xe *3
W e *3
Xe *2
(1.147)
e *3
>
>X (e
@ e
e )@ e
e *3 X (e *3 e *2 e *2 e *3 )
*
2
e *2
*
3
X
*
3
e *2
e *2
*
3
(1.148)
Then, the tensor components of W in the basis formed by the orthonormal basis e 1* , e *2 ,
e *3 , are given by:
Wij*
0
0 0
0 0 X
0 X 0
(1.149)
In Figure 1.17 we can see these components and the axial vector representation. Note that
if we take any basis that is formed just by rotation along the e 1* -axis, the components of
W in this new basis will be the same as those provided in (1.149).
x3
x2
&
w
Xe 1*
e *2
Wij*
0
0 0
0 0 X
0 X 0
e 1*
X
x1
X
e *3
Figure 1.17: Antisymmetric tensor components in the space defined by the axial vector.
1.5.2.2.3
Any arbitrary second-order tensor A can be split additively into a symmetric and an
antisymmetric part:
1
1
(A A T ) (A A T )
2
2
A sym
A sym A skew
(1.150)
A skew
A ijsym
A ijskew
1
( A ij A ji )
2
(1.151)
B A
sym
>
T
1 T
1 T
A B A AT B A
A B A A T BT A
2
1
A T B B T A A T B sym A
2
>
(1.152)
1 TENSORS
41
V1 j W1 j V 2 j W2 j V 3 j W3 j
V31W31
V21W21
V11W11
V32 W32
V 22 W22
V12 W12
V33W33
V23W23
V13W13
Taking into account the characteristics of a symmetric and an antisymmetric tensor, i.e.
V12 V 21 , V 31 V13 , V 32 V 23 , and W11 W22 W33 0 , W21 W12 , W31 W13 ,
W32 W23 , the equation above becomes:
:W
&
&
&
&
&
&
&
symmetric tensor
&
& &
&
M Q M M Q sym M
b)
( A sym A skew ) : (B sym B skew )
A :B
skew
sym
skew
A sym : B sym
A sym
A skew
: B
: B skew
: B
A
0
A sym : B sym
A : B skew
A skew : B skew
&
Problem 1.17:
Let T be
an arbitrary second-order tensor, and n be a vector. Check if the
&
&
relationship n T T n is valid.
Solution:
&
n T
n i e i Tkl (e k e l )
n i Tkl E ik e l
n k Tkl e l
&
T n
and
Tlk (e l e k ) n i e i
n i Tlk E ki e l
n k Tlk e l
(n1 Tl1 n 2 Tl 2 n 3 Tl 3 )e l
&
&
n T z T n
42
&
T sym n holds.
&
Problem 1.18: Obtain the axial vector w associated with the antisymmetric tensor
& &
( x a ) skew .
&
Solution: Let z be an arbitrary vector, it then holds that:
& &
& & &
( x a ) skew z w z
&
& &
where w is the axial vector associated with ( x a ) skew . Using the definition of an
antisymmetric tensor:
>
& &
1 & & & &
1 & &
>x a a x @
( x a) ( x a)T
2
2
& & skew & & &
and by replacing it with ( x a ) z w z , we obtain:
1 & & & & & & &
>x a a x @ z w z >x& a& a& x& @ z& 2w& z&
2
& & & & & &
& &
By using the equation >x a a x @ z z ( x a ) , (see Eq. (1.105)), the above equation
& &
( x a ) skew
becomes:
& &
& &
&
&
z ( x a ) (a x ) z
& &
2w z
1.5.2.3
& &
1 & &
(a x ) is the axial vector associated with ( x a ) skew
2
&
Let A be a second-order tensor and a , b be arbitrary vectors then there is then a unique
tensor cof(A ) , known as the cofactor of A , as we can see below:
&
& &
&
cof( A ) (a b) ( A a) ( A b)
(1.153)
>cof (A )@T
(1.154)
>adj(A)@T
adj( A T )
(1.155)
The components of cof(A ) are obtained by expressing the equation (1.153) in terms of its
components, i.e.:
>cof(A)@it . tpr a p b r
. ijk A jp a p A kr b r
. ijk A jp A kr
(1.156)
By multiplying both sides of the equation by . qpr and by also considering that
. tpr . qpr 2E tq , we can conclude that:
>cof(A)@it . tpr
. ijk A jp A kr
2E tq
>cof( A )@iq
1.5.2.4
1
. ijk . qpr A jp A kr
2
Tensor Trace
. ijk . qpr A jp A kr
(1.157)
1 TENSORS
Tr (e i e j ) e i e j
43
E ij
(1.158)
Tr ( A ij e i e j )
A ij (e i e j )
A ij Tr (e i e j )
A ij E ij
A 11 A 22 A 33
& &
And, the trace of the dyad (u v ) can be evaluated as:
& &
& &
Tr (u v ) Tr (u v ) u i v j Tr (e i e j ) u i v j (e i e j ) u i v j E ij
& &
u1 v 1 u 2 v 2 u 3 v 3 u v
A ii
ui v i
(1.159)
(1.160)
NOTE: As we will show later, the tensor trace is an invariant, i.e. it is independent of the
coordinate system.
Let A , B be arbitrary tensors, then:
Tr A T
Tr A
(1.161)
(1.162)
TrA B Tr A Tr B
>A 11 B 11 A 22 B 22 A 33 B 33 @ A 11 A 22 A 33 B 11 B 22 B 33
(1.163)
>
Tr ( A ij e i e j ) (B lm e l e m )
A ij B lm E jl Tr (e i e m )
>
A B
A il B li
E im
(1.164)
Tr B A
and, the double scalar product ( : ) can be expressed in trace terms as:
A :B
A ij B ij
A kj B lj E ik E il
A ik B il E jk E jl
A kj B lj E kl
A B E
ik
il kl
( A
BT )
kl
A B
T
kk
Tr ( A B )
T
( AT B )
kl
(1.165)
B kk
Tr ( A T B)
T
(1.166)
A ii
>Tr (A )@
TrA A
2
Tr ( A ) Tr ( A )
Tr A
A ii A jj
A il A li ; TrA A A Tr A 3
(1.167)
A ij A jk A ki
44
T T
m T
Solution:
T T T T
m T
T m
and
Tr T T
TT TT TT
Tr T m .
T m
For the second demonstration we can use the trace property Tr T T Tr T , thus:
Tr T T
Tr T m
Tr T m
1.5.2.5
Particular Tensors
1.5.2.5.1
Unit Tensors
The second-order unit tensor, also called the identity tensor, is defined as:
1 E ij e i e j
e i e i
1 e i e j
(1.168)
where 1 is the matrix with the components of tensor 1 . E ij is the Kronecker delta symbol
defined in (1.48).
I ijk" e i e j e k e "
(1.169)
I 11 E i" E jk e i e j e k e "
I ijk"
e i e j e k e "
(1.170)
I 1 1 E ij E k" e i e j e k e "
I ijk" e i e j e k e "
(1.171)
Taking into account the fourth-order unit tensors defined above, it holds that:
I:A
(1.172)
and
I:A
(1.173)
AT
and
I :A
(1.174)
Tr ( A )1
1
o I ijk"
11 11 incomponents
2
1
E ik E j" E i" E jk
2
(1.175)
The property of the tensor product is presented below. Consider a second-order unit
tensor, 1 E ij e i e j . Then, the tensor product can be defined as:
1 TENSORS
11
45
ij e i
(1.176)
I 11 E ik E j" e i e j e k e "
(1.177)
11
ij e i
or
I 11 E i" E jk e i e j e k e "
(1.178)
(1.179)
1
11 11
2
I skew
incomponents
o
skew
I ijk
"
&
1
E ik E j" E i" E jk
2
(1.180)
With a second-order tensor A and a vector b , the following relationships are valid:
&
&
b 1 b
I:A
A :1
Tr (A )
A :1
3
A :1
I sym : A
A ii
Tr A
Tr A
A sym
(1.181)
Tr A A A il A li
Tr A A A A ij A jk A ki
Levi-Civita Pseudo-Tensor
The Levi-Civita Pseudo-Tensor, also known as the Permutation Tensor, is a third-order pseudotensor and is denoted by:
. . ijk e i e j e k
(1.182)
which is not a true tensor in the strict meaning of the word, and whose components . ijk
were defined in (1.55), the permutation symbol.
1.5.2.6
Determinant of a Tensor
. ijk A 1i A 2 j A 3k
. ijk A i1 A j 2 A k 3
AT
(1.183)
46
A 12
A 13
A 21 A 22 A 23
A 31 A 32 A 33
A 11 ( A 22 A 33 A 23 A 32 ) A 21 ( A 12 A 33 A 13 A 32 ) A 31 ( A 12 A 23 A 13 A 22 )
A
A 11 (.1 jk A j 2 A k 3 ) A 21 (. 2 jk A j 2 A k 3 ) A 31 (. 3 jk A j 2 A k 3 )
.1 jk A 11 A j 2 A k 3 . 2 jk A 21 A j 2 A k 3 . 3 jk A 31 A j 2 A k 3
. ijk A i1 A j 2 A k 3
(1.184)
(1.185)
(1.186)
where B is a scalar
(1.187)
If all elements of a row or column equal zero, the determinant is also zero.
If you multiply all the elements of a row or column by a constant c (scalar), the
determinant is c A .
If two or more rows (or column) are linearly dependent, the determinant is zero.
E rt
E jt
E kt
E rp
E jp
E kp
E rq
E jq
E kq
(1.188)
(1.189)
E rt E jp E kq E rp E jq E kt E rq E jt E kp E rq E jp E kt E jq E kp E rt E kq E jt E rp
Then, by substituting (1.189) into (1.188) we can obtain:
A . tpq
A t1 A p 2 A q 3 A p1 A q 2 A t 3 A q1 A t 2 A p 3 A q1 A p 2 A t 3 A t1 A q 2 A p 3 A p1 A t 2 A q 3
A t1 .1 jk A pj A qk A t 2 . 2 jk A pj A qk A t 3 . 3 jk A pj A qk
. rjk A rt A jp A kq . rjk A tr A pj A qk
1
. rjk . tpq A rt A jp A kq .
6
1 TENSORS
47
Starting with the definition A . tpq . rjk A rt A jp A kq , (see Problem 1.21), and by multiplying
both sides of the equation by . tpq , we obtain:
A . tpq . tpq
. rjk . tpq A rt A jp A kq
Using the property defined in expression (1.62), we obtain
. tpq . tpq E tt E pp E tp E tp E tt E pp E tt 6 . Then, the relationship (1.190) becomes:
A
(1.190)
1
. rjk . tpq A rt A jp A kq
6
&
&
& &
det N1 Ba b
& &
N 3 N 2B a b
(1.191)
NB a j b 2 a k b 3 E i1 NB 2 a i a k b 1b 3 E j 2 NB 2 a i a j b 1b 2 E k 3 B 3 a i a j a k b 1b 2 b 3
2
N 3 .123
N3 ,
& &
N 2 B (.12 k a k b 3 . 1 j 3 a j b 2 . i 23 a i b1 ) N 2 B (a 3b 3 a 2 b 2 a1b1 ) N 2 B (a b)
. ijk a i a k b1b 3 E j 2
. i 2 k a i a k b 1b 3
a 1 a 3 b 1b 3 a 3 a 1b 1b 3
&
& &
& &
det 1 Ba b 1 B a b
& &
Then, it is simple to prove that det Ba b 0 , since
& & &
& &
det Ba b B 3 . ijk a i a j a k b1b 2 b 3 B 3b1b 2 b 3 >a (a a)@ 0
>
>
48
& &
& &
det 1 B a b B b a
& &
& &
& &
1 B ab B ab B2 ab
& &
& & 2
1 B 2 a b B a b
& &
(1.194)
>
>
>
&
&
&
( A a) ( A b) ( A c )
>
(1.197)
To achieve this goal we start with the definition of the scalar triple product given in (1.69),
& & &
i.e. a b c . ijk a i b j c k , and by multiplying both sides of this equation by the
determinant of A we obtain:
. ijk a i b j c k A
1.5.2.7
. ijk a i b j c k A
(1.198)
. pqr A pi A qj A rk , thus:
>
@ >
(1.199)
Inverse of a Tensor
A 1 A 1
(1.200)
A ik1 A kj
(1.201)
Or in indicial notation:
if A z 0 A ij1 A ik A kj1
E ij
To obtain the inverse of a tensor we start from the definition of the adjugate tensor given
&
&
& &
in (1.153), i.e. adj( A T ) (a b) ( A a) ( A b) . Then by applying the dot product
&
^>adj(A)@
(a b)` d
&
&
&
(1.202)
thus,
1 TENSORS
^>adj(A)@
(a b)` d
&
&
&
49
1
&
&
& &
A a b A 1 d
(1.203)
&
& &
Denoted by p (a b) , the above equation (1.203) can be rearranged as follows:
(1.204)
>adj(A)@
A 1
1
>adj(A)@
A
1
>cof(A)@T
A
(1.205)
1 1
CA 1
det ( A 1 )
A A 1
B 1 A 1
A
1
(1.206)
A 1
>det( A )@1
(1.207)
Next, we prove the relation adj( A B) adj(B) adj( A ) holds. To do this, we take the
definition of the inverse of a tensor given in (1.205) as a starting point:
B 1 A 1
A B A B
adj( A B)
1
(1.208)
A 11
A 12
A 31
A 32
A A 21 A 22
A 13
A 23
A 33
(1.209)
We define the matrix M , where the component Mij is the determinant of the resulting
matrix by removing the i th row and the j th column, i.e.:
50
A 22
A 32
A
M 12
A 32
A 12
A 22
A 23
A 33
A 13
A 21
A 31
A 11
A 23
A 33
A 13
A 21
A 31
A 11
A 22
A 32
A 12
A 33
A 13
A 31
A 11
A 33
A 13
A 23
A 21
A 23
A 31
A 11
A 21
A 32
A 12
A 22
(1.210)
(1.211)
A 1
>cof (A )@T
(1.212)
adj(A )
(1.213)
& &
0 n z 0 . Secondly, we show that, if
0.
>
&
&
&
( A f ) ( A g) ( A h)
>
&
&
&
&
&
&
&
B ( A f ) C ( A g) H ( A h) 0 A Bf Cg Hh
&
&
&
&
&
& &
0 A n 0
&
&
1 TENSORS
51
z0
1.5.2.8
Orthogonal Tensors
QT Q 1
Q ik Q jk
Q ki Q kj
E ij
(1.214)
QT
If Q
(1.215)
1
(1.216)
( A B ) 1
vectors becomes:
&
&
~
b Q b
&
&
~
And the dot product between these new vectors ( ~a ) and ( b ) is given by:
& &
&
& & &
&
&
~ ~
a b (Q a) (Q b) a Q T Q b a b
&
&
~
a Qa
~~
ai b i
(Q ik a k )(Q ij b j ) a k Q ik Q ij b j
ak b k
(1.218)
(1.219)
E kj
& &
& ~&
~ ~
So, it is also true when ~a b , thus a
a
&
~
a
& &
aa
& 2
a . Therefore, we can conclude
that in an orthogonal transformation, the magnitude vectors and the angle between them
are maintained, (see Figure 1.18).
52
&
b
T
&
~
b
&
a
&
~
a
&
~
a
&
a
&
~
b
&
b
& &
~ ~ & &
ab ab
Indicial notation
Matrix notation
(1.220)
x T x ! 0
xT T x ! 0
&
for all vectors x z 0 . Conversely, a tensor is said to be negative definite when these notations
x i Tij x j ! 0
hold:
Tensorial notation
Indicial notation
Matrix notation
x i Tij x j 0
xT T x 0
x T x 0
& &
for all vectors x z 0 .
(1.221)
&
Tij
wx j
wx i
x j Tij x i
wx k
wx k
Tij E ik x j Tij x i E jk
Tkj x j Tik x i
Tki Tik x i
(1.222)
2 T sym x
w 2B
wx wx
2 T sym
(1.223)
Remember that it is also true that x T x x T sym x , therefore if the symmetric part of
a tensor is positive definite the tensor is too.
NOTE: As we will demonstrate later, the eigenvalues must be positive for T to be
positive definite. The proof is in the subsection Spectral Representation of Tensors.
Problem 1.25: Let F be an arbitrary second-order tensor. Show that the resulting tensors
C F T F and b F F T are symmetric tensors and semi-positive definite tensors. Also check in
what condition are C and b positive definite tensors.
Solution: Symmetry:
1 TENSORS
CT
b
(F T F )T
(F F )
T T
53
F T (F T )T
T T
(F )
FT F
F F
F x F x
( F x ) ( F x )
F x
Or in indicial notation:
x i C ij x j
x ( F F T ) x
t0
x i ( Fki Fkj ) x j
( Fki x i )( Fkj x j )
Fki x i
x i bij x j
x F F T x
( F T x ) ( F T x )
2
F T x t 0
x i ( Fik F jk ) x j
( Fik x i )( F jk x j )
t0
Fik x i
t0
that x C x
F x
C
F T F and b
where
U S BT
(1.224)
Note that, depending on the value of B , we have an infinite number of possibilities for
representing S . But, if Tr ( T UT ) Tr (U T T ) 0 , the additive decomposition is unique.
From the relationship in (1.224), we can evaluate the value of B as follows:
S TT
BT T T U T T Tr (S T T ) BTr ( T T T )
Tr (U T T ) BTr ( T T T )
Tr (S T T )
Tr ( T T T )
(1.225)
(1.226)
Tr (S T T )
Tr ( T T T )
Tr (S 1)
Tr (1 1)
Tr (S )
Tr (1)
Tr (S )
3
(1.227)
S
Tr (S )
1 { S dev
3
(1.228)
Thus:
S
Tr (S )
1 S dev
3
S sph S dev
(1.229)
54
Tr (S )
1 is the spherical part of the tensor S , and S dev
3
NOTE: S sph
S
Tr (S )
1 is
3
>
1
Tr S (S S T ) T
2
Tr (S T T )
Tr ( T T T )
1
(S S T ) then B can be evaluated as follows:
2
>
1
Tr (S S T ) (S S T ) T
4
1
2
(1.230)
1
(S S T ) . Then we obtain S
2
1
1
(S S T ) (S S T ) S sym S skew
2
2
(1.231)
which is the same as the equation obtained in (1.150) in which we split the tensor into
symmetric and antisymmetric parts.
Problem 1.26: Find a fourth-order tensor P so that P : A A dev , where A is a secondorder tensor.
Solution: Taking into account the additive decomposition into spherical and deviatoric parts,
we obtain:
A
A sph A dev
Tr ( A )
1 A dev
3
A dev
A
Tr ( A )
1
3
Referring to the definition of fourth-order unit tensors seen in (1.172), and (1.174), where
the relations I : A Tr ( A )1 and I : A A hold, we can now state:
A dev
A
Tr ( A )
1
1
1 I : A I : A I I: A
3
3
3
I 1 1 : A
3
1
I 1 1
3
1.5.3
The tensor components depend on the coordinate system, so, if the coordinate system is
changed due to a rotation so do the tensor components. The tensor components between
these coordinate systems are interrelated to each other by the component transformation
law, which is defined below, (see Figure 1.19).
Consider a Cartesian coordinate system x1 , x 2 , x3 formed by the orthogonal basis
e 1 , e 2 , e 3 , (see Figure 1.20). In this system, an arbitrary vector v& is represented by its
components as follows:
&
v
v i e i
v 1 e 1 v 2 e 2 v 3 e 3
(1.232)
1 TENSORS
55
TENSORS
Mathematical interpretation of physical concepts
(Independent of the coordinate system)
COMPONENTS
Representation of a tensor in
a coordinate system
COMPONENT
TRANSFORMATION
LAW
COORDINATE SYSTEM
I
COORDINATE SYSTEM
II
H1
xc3
xc1
2
ec
3
ec
e 3
1
ec
C1
x2
e 2
e 1
B1
x1
56
vi
v1
v
2
v 3
(1.233)
Now consider a new coordinate system x1c , x 2c , x3c represented by the orthogonal basis
e 1c , e c2 , e c3 , (see Figure 1.20). In this new system, the vector v& is represented by v cj e cj . As
mentioned before, a tensor is independent of the adopted system, so:
&
v
v ck e ck
v j e j
(1.234)
To obtain the components of a tensor in a given system one need only make the dot
product between the tensor and the system basis:
v ck e ck e ci
( v j e j ) e ci
v ck E ki
( v j e j ) e ci
(1.235)
( v 1 e 1 v 2 e 2 v 3 e 3 ) e ci
v ci
Or in matrix form:
( v 1e 1 v 2 e 2 v 3 e 3 ) e 1c
( v 1e1 v 2 e 2 v 3 e 3 ) e c2
( v 1 e 1 v 2 e 2 v 3 e 3 ) e c3
v 1c
v c
2
v c3
(1.236)
e 1 e 1c
e1 e c2
e 1 e c3
e 2 e 1c
e 2 e c2
e 2 e c3
e 3 e 1c v 1
e 3 e c2 v 2
e 3 e c3 v 3
a ij
e j e ci
e ci e j
(1.237)
Or in indicial notation:
v ci
a ij v j
(1.238)
e 1 e 1c
e1 e c2
e 1 e c3
e 2 e 1c
e 2 e c2
e 2 e c3
a ij { A
e 3 e 1c
e 3 e c2
e 3 e c3
a11
a
21
a 31
e 1c e 1
e c2 e1
e c3 e 1
a12
a 22
a 32
a13
a 23
a33
e 1c e 2
e c2 e 2
e c3 e 2
e 1c e 3
e c2 e 3
e c3 e 3
(1.239)
The matrix ( A ) is not symmetric, i.e. A z A T . With reference to the scalar product
e ci e j e ci e j cosxic , x j cosxic , x j , (see equation (1.4)), the relationship in (1.237) is
expressed by means of the direction cosines as:
1 TENSORS
v 1c
v c
2
v c3
,
vc
57
(1.240)
v c Av
The direction cosines of a vector are those of the angles between the vector and the three
coordinate axes. According to Figure 1.20 we can verify that cos B 1 cosxc1 , x1 ,
cos C 1 cos xc1 , x 2 and cos H 1 cos xc1 , x3 .
i . Now, we can project the
In the equation (1.235) we have projected the vector onto ec
vector onto e i :
v k e k e i
v cj e cj e i
v k E ki
v cj a ji
vi
v cj a ji
A T vc
(1.241)
a ji e cj
(1.242)
A 1 v c
(1.243)
and by comparing the equations (1.243) with (1.241) we can conclude that the matrix A is
an orthogonal matrix, i.e.:
A 1
AT
Indicial
A T A 1 notation
o a ki a kj
E ij
(1.244)
Second-order tensor
Consider a coordinate system formed by the orthogonal basis e i then, how the basis
i . This is
changes from the e i system to a new one represented by the orthogonal basis ec
Tkl e k e l
Tkl a ik e ci a jl e cj
Tkl a ik a jl e ci e cj
Tijc e ci e cj
(1.245)
Then, the transformation law of the components between systems for a second-order
tensor is given by:
Tijc
Tkl a ik a jl
a ik Tkl a jl
Matrix
form
o
T c A T AT
(1.246)
Third-order tensor
A third-order tensor ( S ) can be shown in two systems represented by orthogonal bases e i
i as follows:
and ec
58
S lmn e l e m e n
S lmn ail e ci a jm e cj a kn e ck
S lmn ail a jm a kn e ci e cj e ck
S cijk e ci e cj e ck
(1.247)
i ) are:
In conclusion the components of the third-order tensor in the new basis ( ec
S cijk
S lmn a il a jm a kn
(1.248)
The following table summarizes the transformation law of the components according to
the tensor rank:
rank
to
from x1 , x 2 , x 3 o
x1c , x 2c , x3c
to
from x1c , x 2c , x 3c o
x1 , x 2 , x3
0 (scalar)
Oc O
O Oc
S ic
1 (vector)
S ijc
c
S ijk
3
4
c
S ijkl
a ij S j
Si
a ik a jl S kl
S ij
a il a jm a kn S lmn
S ijk
a im a jn a kp a lq S mnpq
S ijkl
a ji S cj
a ki a lj S klc
(1.249)
c
a li a mj a nk S lmn
c
a mi a nj a pk a ql S mnpq
where the components of T and A are shown, respectively, as Tij and a ij . Afterwards,
given that a ij are the components of the transformation matrix, represent graphically the
components of the tensors T and Tc on both systems.
Solution: The expression T c A T A T in symbolic notation is given by:
c (e a e b )
Tab
a rs (e r e s ) T pq (e p e q ) a kl (e l e k )
a rs T pq a kl E sp E ql (e r e k )
a rp T pq a kq (e r e k )
To obtain the components of T c one only need make the double scalar product with the
basis (e i e j ) , the result of which is:
c (e a e b ) : (e i e j )
Tab
c E ai E bj
Tab
Tijc
a rp T pq a kq (e r e k ) : (e i e j )
a rp T pq a kq E ri E kj
a ip T pq a jq
1 TENSORS
59
T c A T AT
xc3
Tc33
Tc23
x3
xc3
T33
Tc13
Tc31
T13
T23
x1
Tc22
xc2
Tc21
T22
T31
T11
Tc12
Tc11
xc2
T32
Tc32
T12
T21
xc1
x2
T AT T c A
Figure 1.21: Transformation law of the second-order tensor components.
xc1
Tr ( T )
Tii
II T
1
2
I T Tr ( T 2 )
2
III T
det ( T )
Tr ( T c T c)
Tc : Tc
Tijc Tijc
( a ik a jl Tkl )( a ip a jq T pq )
a ik a ip a jl a jq Tkl T pq
E kp
E lq
T pl T pl
T : T Tr ( T T )
Tr ( T 2 )
c)
det ( T c)
det ( T c)
det (A T A T )
det ( T )
60
1
B 1 B T
BA
X cc
X
T
A B
(BA )
CBA
(CBA ) T
AT BT CT
C 1
X ccc
CT
(1.250)
(1.251)
v cc Bv c
(1.252)
v c B T v cc
(1.253)
v cc BA v
(1.254)
Now, starting with the system x1c , x c2 , x3c , the components of the vector in the system
x1cc, x c2c , x3cc are given by:
and the inverse transformation is:
(1.255)
A T B T v cc
This equation could have been obtained by using equation (1.254), i.e.:
BA 1 v cc BA 1 BA v
BA 1 v cc
A 1B 1v cc A T B T v cc
(1.256)
Then, it is easy to find the components of the vector in the coordinate system x1ccc, x 2ccc, x3ccc ,
(see Figure 1.22):
v ccc CBA v
inverse
form
o
A T B T C T v ccc
(1.257)
1 TENSORS
1.5.3.1
61
B y cy
xc
yc
B xcy
B xcy
B y cy
B y cx
B xcx
B y cx
B
S
B
2
S
B
2
(1.258)
B sin(B ) ,
2
(1.259)
sin(B ) cos(B )
(1.260)
Another way to prove (1.260) is by considering the vector position of the point P in both
systems, (see Figure 1.24).
Moreover, in view of Figure 1.24, said coordinates are interrelated as shown below:
x cP
y cP
x cP
S
x P cos(C ) y P cos C
yc
2
x P cos(B ) y P cos(C )
xc
P
y cP
S
x P cos(B ) y P cos B
2
S
x P cos B y P cos(B )
(1.261)
x P cos(B ) y P sin(B )
x P sin(B) y P cos(B )
Or in matrix form:
x cP
yc
P
Inverse
xP
cos(B) sin(B ) x P transforma
tion
sin(B ) cos(B ) y o y
P
P
cos(B ) sin(B )
sin(B ) cos(B )
1
x cP
yc
P
(1.262)
62
xP
y
P
cos(B ) sin(B ) x cP
sin(B ) cos(B ) y c
(1.263)
yPc
xc
&
r
xcP
C
ycP
yP
xP
xP
co
D)
s(
co
yc
)
(D
sin
yP
xP
P
yP
)
(D
sin
D)
s(
Problem 1.29: Find the transformation matrix between the systems: x, y , z and xccc, y ccc, zccc .
These systems are represented in Figure 1.25.
z
zc
z cc
z ccc
y ccc
yc
y cc
B
x
xc
H
x ccc
x cc
1 TENSORS
63
Solution: The coordinate system xccc, y ccc, zccc can be obtained by different combinations of
rotations as follows:
i
from x, y , z to xc, y c, zc
yc
with 0 d B d 360
cos B sin B 0
sin B cos B 0
0
0
1
xc
zc
z cc
B
yc
y cc
B
B
cos C
0
sin C
with 0 d C d 180
y
z
xc
zc
z cc
xc
x cc
0 sin C
1
0
0 cos C
x cc
zc
z cc
z ccc
y ccc
yc
y cc
D
x
xc
sin H
cos H
0
with 0 d H d 360
H
x ccc
x cc
cos H
sin H
0
0
1
64
The transformation matrix from ( x, y , z ) to ( xccc, y ccc, z ccc ), (see Figure 1.22), is given by:
D CBA
After multiplying the matrices, we obtain:
(sin B cos C cos H cos B sin H ) sin C cos H
(cos B cos C cos H sin B sin H )
D ( cos B cos C sin H sin B cos H ) ( sin B cos C sin H cos B cos H ) sin C sin H
cos B sin C
sin B sin C
cos C
The angles B, C , H are known as Euler angles and were introduced by Leonhard Euler to
describe the orientation of a rigid body motion.
Problem 1.30: Let T be a second-order tensor whose components in the Cartesian system
x1 , x 2 , x3 are given by:
T ij
Tij
3 1 0
1 3 0
0
0 1
Given that the transformation matrix between two systems, x1 , x 2 , x 3 - x1c , x 2c , x 3c , is:
0
2
2
2
0
2
2
2
2
Obtain the tensor components Tij in the new coordinate system x1c , x 2c , x 3c .
Solution: As defined in equation (1.249), the transformation law for second-order tensor
components is:
Tijc
aik a jl Tkl
>a i k @ >Tk l @
>a @
l j
Thus
Tc
2
2
T c A T AT
0
2
2
2
2
1
0
3 1 0
0 1 3 0 0
0 1
0
1
0
2
2
2
2
0
2
2
2
1 0 0
0 2 0
0 0 4
NOTE: As we can verify in the above example, the components of the tensor T , in the
new basis, have one particular feature, i.e. the off-diagonal terms are equal to zero. The
question now is: Given an arbitrary tensor T , is there a transformation which results in the
1 TENSORS
65
off-diagonal terms being zero? This type of problem is called the eigenvalue and eigenvector
problem.
1.5.4
As we have seen, the scalar product between a second-order tensor T and a vector (or unit
) leads to a vector. In other words, projecting a second-order tensor onto a
vector nc
,
certain direction results in a vector that does not necessarily have the same direction as nc
(see Figure 1.26(a)).
The aim of the eigenvalue and eigenvector problem is to find a direction n , in such a way
&
that the resulting vector, t (n) T n , coincides with it, (see Figure 1.26 (b)).
b) Principal direction.
a) Projection of T onto an
arbitrary plane.
&
t (nc)
T n c
&
t (n )
T n On
nc
n - principal direction of T
x3
O - eigenvalue of T associated
with the direction n .
x2
x1
(1.264)
On i
Tij n j On i
Tij OE ij n j
0i
0i
(1.265)
&
o T O1 n 0
Tensorial
notation
&
The previous set of homogeneous equations only have nontrivial solution, i.e. n z 0 , if and
only if:
det ( T O1) 0 ;
Tij OE ij
(1.266)
The determinant (1.266) is called the characteristic determinant of the tensor T , explicitly given
by:
66
T11 O
T21
T12
T22 O
T13
T23
T31
T32
T33 O
(1.267)
(1.268)
where I T , II T , III T are the principal invariants of T , and are defined in components terms
as:
IT
II T
Tr (T )
Tii
E jk
1
( TrT ) 2 Tr ( T 2 )
2
1
Tr ( Tij e i e j ) Tr ( Tkl e k e l ) Tr Tij e i e j Tkl e k e l
2
1
Tij E ij Tkl E kl Tij Tkl E jk Tr e i e l
2
1
Tii Tkk Tij Tkl E jk E il
2
1
Tii Tkk Tij T ji
Mii Tr>cof( T )@
2
>
^
^
>
III T
det ( T )
Tij
>
@`
II T
III T
(1.269)
. ijk Ti1 T j 2 Tk 3
@`
(1.270)
II T
III T
2
T11 T22 T11 T33 T22 T33 T122 T132 T23
2
2
T11 T22 T33 T12 T13 T23 T13 T12 T23 T12 T33 T23
T11 T132 T22
(1.271)
1 TENSORS
67
O 1
0
0
O2
0
T1
0
0
0
O 2
0
0
T3
0
T2
0
(1.272)
T1 T2 T3 , II T
T1 T2 T2 T3 T1 T3 , III T
T1 T2 T3
(1.273)
whose values must match the values obtained in (1.270), since they are invariant with a
change of basis.
If T is a spherical tensor, i.e. T1
T2
T , it holds that I T2
T3
3 II T , III T
T3.
IW
>
II W
Tr (W 2 )
2
W13
0
0
W12
W12
0
(1.274)
X2
III W
&
& &
2
where X 2 w
w w W23
W132 W122 as defined in (1.132). Then, the characteristic
equation for an antisymmetric tensor is reduced to:
2
O3 O2 I W O II W III W
O3 X 2 O
O O2 X 2
(1.275)
In this case, one eigenvalue is real and equal to zero and the others are imaginary roots:
O2 X 2
1.5.4.1
O2
X 2
O (1, 2 )
rX 1 rX i
(1.276)
O 1n (1)
T n ( 2 )
T n (3)
O 2 n ( 2)
O 1n ( 2 ) n (1)
n (1) T n ( 2 )
O 2 n (1) n ( 2 )
O 3 n (3)
(1.277)
(1.278)
O 2 n (1) n ( 2 )
O 2 n ( 2 ) n (1)
(1.279)
68
O 1 O 2 n (1) n ( 2 )
(1.280)
(1.281)
n (1)
( 2)
n
n (3)
n 1(1)
(2)
n1
n (3)
1
( 3)
n (21)
n (22 )
n (23)
(2)
( 3)
n 3(1)
n 3( 2 )
n (33)
(1.282)
diagonalization
T c A T AT
x3
T3
T33
xc3
xc3
T23
T13
n (3)
T2
(2)
xc2
xc2
T23
T13
n (1)
T22
T12
T12
T1
x2
T11
xc1
x1
xc1
AT T c A
Principal Space
C14 C 24 C 34
where C1 , C 2 , C 3 are the eigenvalues of the second-order tensor C .
;
C13 C 23 C 33
Solution: Any combination of invariants is also an invariant, so, on this basis, we can try to
express the above expressions in terms of their principal invariants.
I C2
C1 C 2 C 3 2
C12 C 22 C 32 2 C1 C 2 C1 C 3 C 2 C 3 C12 C 22 C 32
II C
I C3 3 II C I C 3 III C
C14
I C4 4 II C I C2 4 III C I C 2 II C2
C 24
C 34
I C2 2 II C
1 TENSORS
69
Q E QT
det Q E Q O1
det Q E Q Q O1 Q
det >Q E O1 Q @
det E O1 det
Q
det
Q
0 det E * O1
T
T
det E O1
det Q
det Q
0 det E *ij OE ij
ik E kp Q jp
OE ij
ik E kp Q jp
OQ ik Q jp E kp
>
@
det Q det E OE det Q
det E OE
det Q ik E kp OE kp Q jp
ik
kp
kp
jp
kp
kp
B I
2 S cos T
3 3
B 2S
2 S cos
3
3
B 4S
2S cos
3
3
IT
3
(1.283)
IT
3
where
R
I T2 3 II T
;
3
R
;
3
I T II T
2I 3
III T T ;
3
27
R3
;
27
2T
B arccos
where B is in radians.
(1.284)
0
O2
0
0
0
O 3
B
0
0
cos 3
1 0 0
IT
B 2S
2S 0
0
1
0
cos
0
3
3
3
0 0 1
B 4S
0
0
cos
Spherical part
33
(1.285)
Deviatoric part
where we clearly distinguish the spherical and the deviatoric part of the tensor in the
principal space. Note that, if T is a spherical tensor the following relationship holds
I T2 3 II T , then S 0 .
70
Problem 1.33: Find the principal values and directions of the second-order tensor T ,
where the Cartesian components of T are:
T ij
Tij
3 1 0
1 3 0
0 1
0
T12
T22 O
T13
T23
3 O 1
1 3 O
T31
T32
T33 O
0
0
1 O
>
(1 O ) (3 O ) 2 1
3
O 7 O 14O 8 0
III T
Tr ( Tij )
Tii
1
Tii T jj Tij Tij
2
Tij
T22
T23
T32
T33
7
T11
T13
T31
T33
T11
T12
T21
T22
14
. ijk Ti1 T j 2 Tk 3 8
O3 7O2 14O 8 0
1;
O1 O 2 O 3
1 2 4
O 1O 2 O 2 O 3 O 3 O 1
O2
2;
O3
1 u 2 2 u 4 4 u 1 14
III T O 1 O 2 O 3 8
Thus, we can see that the invariants are the same as those evaluated previously.
Principal directions:
Each eigenvalue, O i , is associated with a corresponding eigenvector, n (i ) . We can use the
equation in (1.265), i.e. ( Tij OE ij ) n j 0 i , to obtain the principal directions.
O1 1
3 O 1
1
1
3 O1
0
0 n1
0 n 2
1 O 1 n 3
0 n1
3 1 1
1 3 1 0 n
2
0 1 1 n 3
0
0
0
0
1 TENSORS
2n1 n 2 0
n1
n1 2n 2 0
0n
3 0
n12 n 22 n 32
nini
n2
71
1
0 n1
3 O 2
1
n
3
O
0
2
2
0
0
1 O 2 n 3
n1 n 2 0 n1 n 2
n1 n 2 0
n
3 0
0 n1
3 2 1
1 3 2
0 n 2
0
0
1 2 n 3
0
0
0
The first two equations are linearly dependent, after which we need an additional equation:
n12 n 22 n 32
nini
1 2n12
1
2
1 n1
n i( 2 )
1
r
2
Thus:
O2
O3
1
0 n1
3 O 3
1
3
O
0 n 2
3
0
0
1 O 3 n 3
n1 n 2 0
n1 n 2
n1 n 2 0
3n
0
3
nini
1
2
n12 n 22 n 32
0 n1
3 4 1
1 3 4
0 n 2
0
0
1 4 n 3
1 2n 22
1 n2
O3
1
2
Then:
4
n i(3)
1
2
1
2
n i(1)
>0
0 r 1@
O2
n i( 2)
1
2
1
2
O3
n i(3)
1
2
1
2
0
0
0
72
NOTE: The tensor components of this problem are the same as those used in Problem
1.30. Additionally, we can verify that the eigenvectors make up the transformation matrix,
A , between the original system, x1 , x 2 , x 3 , and the principal space, x1c , x 2c , x 3c , (see
Problem 1.30).
1.5.5
n (i1)
T2
n (i 2 )
T3
n i(3)
> n
> n
> n
(1)
1
n (21)
n 3(1)
( 2)
1
n (22)
n 3( 2 )
( 3)
1
n (23)
n (33)
@
@
(1.286)
The principal space is formed by the orthogonal basis n (1) , n ( 2) , n (3) , and the tensor
components are represented by their eigenvalues as:
Tijc
T1
0
T2
T c 0
0
0
T3
(1.287)
With reference to the fact that eigenvectors form a transformation matrix, A , so that:
T c A T AT
(1.288)
AT T c A
(1.289)
where
A
n (1)
( 2)
n
n (3)
n 1(1)
(2)
n1
n (3)
1
n (21)
n (22 )
n (23)
n 3(1)
n 3( 2 )
n (33)
(1.290)
T12
T22
T23
T13
T23
T33
n 1(1) n 1( 2 )
(1) ( 2 )
n 2 n 2
n (1) n ( 2 )
3
3
T1 0
A T 0 T2
0 0
n 1(3) T1
n (23) 0
n 3(3) 0
0
0 A
T3
0
T2
0
0 n 1(1)
0 n 1( 2)
T3 n 1(3)
n (21)
n (22 )
n (23)
n (31)
n 3( 2)
n 3(3)
0 0 0
0 0 0
1 0 0
T1A T 0 0 0 A T2 A T 0 1 0 A T3 A T 0 0 0 A
0 0 1
0 0 0
0 0 0
Whereas:
(1.291)
1 TENSORS
n 1(1) n (21)
n (21) n (21)
n 3(1) n (21)
0 0 0
A 0 1 0A
0 0 0
n 1(1) n 1(1)
(1) (1)
n 2 n 1
n (1) n (1)
3 1
n 1( 2 ) n 1( 2 )
( 2) ( 2)
n 2 n 1
n ( 2 ) n ( 2 )
3 1
0 0 0
T
A 0 0 0A
0 0 1
n 1(3) n 1(3)
( 3) ( 3)
n 2 n1
n (3) n (3)
3 1
n 1(3) n (23)
n (23) n (23)
1 0 0
A 0 0 0A
0 0 0
T
n 1( 2 ) n (22 )
n (22 ) n (22 )
n 3( 2 ) n (22 )
n (33) n (23)
73
n 1(1) n 3(1)
(1.292)
n 1(3) n 3(3)
(1.293)
As we can see, the tensor is represented as a linear combination of dyads and the above
representation in tensorial notation becomes:
T
(1.294)
or:
T
n ( a ) n ( a )
a 1
Spectral representation of a
second-order tensor
(1.295)
which is the spectral representation of the tensor. Note that, in the above equation we have to
resort to the summation symbol, because the dummy index appears thrice in the
expression.
NOTE: The spectral representation in (1.295) could easily have been obtained from the
definition of the second-order unit tensor, given in (1.168), i.e. 1 n i n i , which can also
be represented by means of the summation symbol as 1
(a)
n ( a ) . Then, it follows
a 1
that:
T
T 1 T n ( a ) n ( a )
a 1
T n
(a)
n ( a )
a 1
n ( a ) n ( a )
(1.296)
a 1
Ta n ( a ) .
We now consider the orthogonal tensor R . The orthogonal transformation applied to the
. Therefore, it is also possible to
leads to the unit vector n , i.e. n R N
unit vector N
represent the orthogonal tensor R as follows:
3 (a) (a)
N
R R 1 R N
a 1
R N
a 1
(a)
(a )
N
(a )
(a)
N
(1.297)
a 1
The spectral representation is very useful for making algebraic operations with tensors. For
example, tensor power in the principal space can be expressed as:
74
T1n
0
0
T
n
ij
0
T2n
0
0
T3n
(1.298)
Tn
n
a
n ( a ) n ( a )
(1.299)
a 1
Ta n ( a ) n ( a )
(1.300)
a 1
Next, we can show that a positive definite tensor has positive eigenvalues. For this
purpose, we can consider a semi-positive definite tensor, T , by which the condition
&
x T x t 0 holds for all x z 0 . Replacing the tensor by its spectral representation, we
obtain:
x T x t 0
3
x Ta n ( a ) n ( a ) x t 0
a 1
(1.301)
x n ( a ) n ( a ) x t 0
a 1
Ta x n ( a ) n ( a ) x t 0
( x n ) @ t 0
T >
3
(a ) 2
a 1
a 1
!0
(1.302)
T1 ( x n (1) ) 2 T2 ( x n ( 2 ) ) 2 T3 ( x n (3) ) 2 t 0
&
The above expression must hold for all x z 0 . If we take x n (1) , the above equation is
reduced to T1 (n (1) n (1) ) 2 T1 t 0 . The same is true for T2 and T3 . Thus, we have
demonstrated that if a tensor is semi-positive definite, its eigenvalues are greater than or
equal to zero, i.e. T 1 t 0 , T 2 t 0 , T 3 t 0 . Therefore we can conclude that a tensor is positive
definite, i.e. x T x ! 0 , if and only if its eigenvalues are positive and nonzero, i.e. T 1 ! 0 ,
T 2 ! 0 , T 3 ! 0 . Consequently, the positive definite tensor trace is greater than zero. If the
positive definite tensor trace is zero, this implies that the tensor is the zero tensor.
The spectral representation of the fourth-order unit tensor, I , can be obtained starting
from the definition in (1.169), i.e.:
I E ik E j" e i e j e k e "
e i e j e i e j
e b e a e b
(1.303)
a 1 b 1
As I is an isotropic tensor, (see 1.5.8 Isotropic and Anisotropic tensors), then the
representation in (1.303) is also valid in any orthonormal basis, n ( a ) , so:
3
n
a 1 b 1
(a)
n (b ) n ( a ) n (b )
(1.304)
1 TENSORS
75
e i e j e j e i
(1.305)
n (b ) n (b ) n ( a )
(a)
(1.306)
a ,b 1
and
E ij E k" e i e j e k e "
e i e i e k e k
(1.307)
n ( a ) n (b ) n (b )
(a)
(1.308)
a ,b 1
n ( a ) n ( a )
a 1
a ,b 1
a zb
Z V V Z Z ab (O b O a ) n ( a ) n (b)
a ,b 1
a zb
Solution:
It is true that
Z 1 Z n ( a ) n ( a) Z n (a ) n (a ) w& n ( a) n ( a)
3
w n
3
(b )
(a)
a 1
a 1
(a)
a ,b 1
&
&
w n
w n
w n
n
n
w n
w n
n n
n n
(1)
(1)
n ( 2 )
(2)
(1)
( 3)
( 3)
( 2)
( 2)
( 2)
( 2)
( 3)
( 3)
n n
n n
w3 n (3) n ( 2) n ( 2 )
w3
n n
n n
( 3)
( 3)
( 3)
( 3)
( 2)
( 2)
( 3)
w3
w2
w2
Z 32 ,
(1)
( 2)
(1)
( 3)
Z13
Z 31 , w3
Z12
Z 21 ,
the
76
Z Z ab n (a ) n (b)
The terms
ZV
a ,b 1
a zb
ZV
a ,b 1
b 1
azb
O Z
b
ab
O Z
n ( a ) n (b ) n (b ) n (b )
a ,b 1
a zb
ab
n ( a ) n (b )
a ,b 1
azb
and
3
3
V Z O a n ( a ) n ( a )
Z ab n (a ) n (b)
a 1
a ,b 1
azb
O Z
a
ab
n ( a ) n (b )
a ,b 1
azb
Then,
ZV V Z
3
3
(a)
(b )
(a)
(b )
O b Z ab n n
O a Z ab n n
a ,b 1
a ,b 1
a zb
azb
ab (O b
O a ) n ( a ) n (b )
a ,b 1
a zb
1.5.6
Cayley-Hamilton Theorem
The Cayley-Hamilton theorem states that any tensor, T , satisfies its own characteristic
equation, i.e. if the eigenvalues of T satisfy the equation O3 O2 I T O II T III T 0 , so
does the tensor T :
T 3 T 2 I T T II T III T 1 0
(1.309)
One of the applications of the Cayley-Hamilton theorem is to express the power of tensor,
T n , as a combination of T n 1 , T n 2 , T n 3 . For example, T 4 is obtained as:
T 3 T T 2 TI T T T II T III T 1 T
T4
T 3 I T T 2 II T III T T
(1.310)
1 TENSORS
77
(1.311)
>
1
Tr ( T 3 ) I T Tr ( T 2 ) II T Tr ( T )
3
(1.312)
1
3
1
3
3
2
Tr ( T ) Tr ( T ) Tr ( T ) >Tr ( T )@
2
3
2
(1.313)
1
3
1
(1.314)
or in indicial notation
III T
Problem 1.35: Based on the Cayley-Hamilton theorem, find the inverse of a tensor T in
terms of tensor power.
Solution: The Cayley-Hamilton theorem states that:
T 3 T 2 I T T II T III T 1 0
Carrying out the dot product between the previous equation and the tensor T 1 , we
obtain:
T 3 T 1 T 2 T 1 I T T T 1 II T III T 1 T 1
T 2 TI T 1 II T III T T 1
T
1
0 T 1
1
T 2 I T T II T 1
III T
The Cayley-Hamilton theorem also applies to square matrices of order n . Let Anun be a
square n by n matrix. The characteristic determinant is given by:
O1nun A
(1.315)
where 1nun is the identity n by n matrix. Developing the determinant (1.315) we ontain:
On I 1 On 1 I 2 On 2 (1) n I n
(1.316)
(1.317)
By means of the relationship (1.317), we can obtain the inverse of the matrix Anun by
multiplying all the terms by the inverse, A 1 , i.e.:
78
A n A 1 I 1A n 1A 1 I 2 A n 2 A 1 (1) n I n 1A 1 0
then
(1) n 1
A n 1 I 1A n 2 I 2 A n3 (1) n 1 I n 11
In
A 1
(1.318)
(1.319)
det (A ) z 0 .
Problem 1.36: Check the Cayley-Hamilton theorem by using a second-order tensor whose
Cartesian components are given by:
5 0 0
0 2 0
0 0 1
Solution:
The Cayley-Hamilton theorem states that:
T 3 T 2 I T T II T III T 1 0
where I T 5 2 1 8 , II T 10 2 5 17 , III T 10 , and
T
5 3
0
0
0
23
0
0
1
125 0 0
0 8 0
0 0 1
; T
5 2
0
0
0
22
0
0
1
25 0 0
0 4 0
0 0 1
0 0 1
0 0 1
0 0 1
0 0 1
1.5.7
0 0 0
0 0 0
0 0 0
Norms of Tensors
The magnitude (module) of a tensor, also known as the Frobenius norm, is given below:
&
v
& &
v v
v i vi (vector)
(1.320)
T:T
(1.321)
A : A
(1.322)
(1.323)
C :: C
T:T
Tij Tij
I T2 2 II T
(1.324)
1 TENSORS
79
xc2
T2
T:T
I T2 2 II T
T
T1
xc1
T3
xc3
1.5.8
A tensor is called isotropic when its components are the same in any coordinate system,
otherwise the tensor is said to be anisotropic.
i,
Let T and T c represent the tensor components T in the systems e i and ec
respectively, so, the tensor is isotropic if T T c on any arbitrary basis.
v i e i v cj e cj
v ci a ij v j
(1.325)
&
By definition, v is an isotropic tensor if it holds that v i v ci , and this is only possible if
e i e cj , i.e. there is no change of system, or if the tensor is the zero vector, i.e.
&
v i v ci 0 i . Then, the unique isotropic first-order tensor is the zero vector 0 .
a ik a jl E kl
a ik a jk
AA T 1
E ij
(1.326)
An immediate observation of the isotropy of unit tensor 1 is that any spherical tensor
( B1 ) is also an isotropic tensor. So, if a second-order tensor is isotropic it is spherical and
vice versa.
Isotropic third-order tensor
An example of a third-order isotropic tensor is the Levi-Civita pseudo-tensor, defined in
(1.182), which is not a real tensor in the strict meaning of the word. With reference to
the transformation law for the third-order tensor components, (see equation (1.248)), we
can conclude that:
80
.cijk
a il a jm a kn . lmn
A .
, ijk
(1.327)
I ijkl
I ijkl
E ik E jl
I ijkl
E il E jk
(1.328)
D
D
D ijkl
a 0 1 1 a1 11 a 2 11
(1.329)
a 0 E ij E kl a1E ik E jl a 2 E il E jk
Problem 1.37: Let C be a fourth-order tensor, whose components are given by:
C ijkl ME ij E kl N E ik E jl E il E jk
where M , N are constant real numbers. Show that C is an isotropic tensor.
Solution:
Applying the transformation law for fourth-order tensor components:
C cijkl
a im a jn a kp a lq C mnpq
Ma in a jn a kq a lq N a ip a jq a kp a lq a iq a jn a kn a lq
ME ij E kl N E ik E jl E il E jk
C ijkl
1.5.9
Coaxial Tensors
Two arbitrary second-order tensors, T and S , are coaxial tensors if they have the same
eigenvectors. It is easy to show that if two tensors are coaxial, this means the dot product
between them is commutative, and vice versa, i.e.:
T S S T
if
S, T are coaxial
(1.330)
If T and S are coaxial as well as symmetric tensors, the spectral representations of these
tensors are given by:
T
T
a 1
n ( a ) n ( a )
n ( a ) n ( a )
(1.331)
a 1
An immediate result of (1.330) is that the tensor S and its inverse S 1 are coaxial tensors:
1 TENSORS
S 1 S S S 1
S
n ( a ) n ( a )
S 1
a 1
where S a ,
81
a 1
(1.332)
1 (a ) (a )
n n
1
, are the eigenvalues of S and S 1 , respectively.
Sa
If S and T are coaxial symmetric tensors, the resulting tensor ( S T ) becomes another
symmetric tensor. To prove this we start from the definition of coaxial tensors:
T S S T
T S S T
T S ( T S) T
0 2( T S ) skew
0 (1.333)
Then, if the antisymmetric part of a tensor is a zero tensor, it follows that this tensor is
symmetric:
( T S ) skew
( T S ) { ( T S ) sym
(1.334)
&
f (N)
&
&
f (N) n O (n ) n z 0 ,
( a ) , we can obtain:
since det ( F ) z 0 . After that, given an orthonormal basis N
F 1 F
(a)
(a)
N
a 1
F 1 F
(a)
(a )
N
a 1
O n
a
(a)
F N
(a)
(a)
N
(1.335)
a 1
(a )
N
a 1
( 2)
F N
(1)
n
n ( 2 )
( 3)
N
(1)
F N
& (1 )
f (N )
& ( 1)
f (N ) n (1)
( 2)
F N
& (2)
f (N )
& (2)
f (N ) n ( 2 )
O 2 n ( 2 )
( 3)
F N
& ( 3)
f (N )
& ( 3)
f (N ) n (3)
O 3n (3)
( 2)
N
n (3)
& (3)
f (N )
(1)
F N
(1)
N
( 3)
F N
(a) .
Figure 1.29: Projecting F onto N
O 1n (1)
82
&
&
&
&
&
&
(2)
(2)
(3)
( 3)
( 1)
O n
a
(a )
(a)
N
a 1
O R N
a
(a)
(a)
N
a 1
R U
O N
a
(a)
R U
(a)
N
(1.336)
a 1
U R
3
O N
a
(a )
F
( a ) . Note that U is a symmetric
N
a 1
(a) N
( a ) is also
tensor, i.e. U UT . This condition is easily verified by the fact that N
(a) N
( a ) R T n ( a ) n ( a ) R , we obtain:
symmetric. Now considering that n ( a ) R N
3
O n
a
(a)
O n
(a )
N
a 1
(a)
n ( a ) R
a 1
V R
( a ) n ( a ) R
V R
an
(1.337)
a 1
F R
O n
a
(a)
n ( a ) . By
a 1
comparing the spectral representation of U with V , we can conclude that they have the
same eigenvalues but different eigenvectors, and they are related by n ( a ) R N ( a ) .
With reference to the above considerations, we can define the polar decomposition:
F
R U V R
Polar Decomposition
(1.338)
F T R U (R T F ) T U UT U U 2
U r FT F
r C
FT
V R F T
V (F R T )T
V VT
V2
(1.339)
V R and F T , we obtain:
r F FT
r b
(1.340)
Since det ( F ) z 0 , the tensors C and b are positive definite symmetric tensors, (see
Problem 1.25), which implies that the eigenvalues of C and b are all real and positive.
However, up to now, det ( F ) z 0 is the only restriction imposed on the tensor F .
Therefore, we have the following possibilities:
If det ( F ) ! 0
In this scenario, we have det ( F ) det (R )det(U) det ( V )det(R ) ! 0 , which results in the
following cases:
1 TENSORS
83
U
,
V
Positive
definite
tensors
If det ( F ) 0
In this situation, we have det ( F ) det (R )det (U) det ( V )det (R ) 0 , which give us the
following cases:
R Proper orthogonal tensor
R Improper orthogonal tensor
or
NOTE: In Chapter 2 we will work with some special tensors where F is a nonsingular
tensor, det ( F ) z 0 , and det ( F ) ! 0 . U and V are positive definite tensors and R is a
rotation tensor, i.e. a proper orthogonal tensor.
wA ij
(e i e j e k e l ) E ik E jl (e i e j e k e l ) I
wA kl
(1.341)
wA kk
(e i e j ) E ki E kj (e i e j ) E ij (e i e j ) 1
wA ij
(1.342)
The derivative of the tensor trace squared with respect to the tensor is given by:
w>Tr ( A )@
wA
2 Tr ( A )
w>Tr ( A )@
2 Tr ( A )1
wA
(1.343)
And, the derivative of the trace of the tensor squared with respect to tensor is given by:
>
w Tr ( A 2 )
wA
w ( A sr )
w( A rs )
A sr
A rs
(e i e j )
wA ij
wA ij
w ( A sr A rs )
(e i e j )
wA ij
>A
rs
E si E rj A sr E ri E sj @ (e i e j )
>A
ji
A ji (e i e j )
(1.344)
2 A ji (e i e j ) 2A T
>
w Tr ( A 3 )
wA
(1.345)
3( A 2 ) T
2 Tr (C)1 ,
>
w Tr (C 2 )
wC
2C T
2C ,
>
w Tr (C 3 )
wC
3(C 2 ) T
Moreover, we can say that the derivative of the Frobenius norm of C is given by:
3C 2 .
84
wC
wC
w Tr (C C T )
wC
w C:C
wC
1
Tr (C 2 )
2
>
w Tr (C 2 )
wC
>
1
Tr (C 2 )
2
@ >Tr(C )@,
1
2
(1.346)
1
2 2C
or:
wC
wC
C
C
Tr (C )
(1.347)
w n i C ij n j
wn j
wn i
C ij n j n i C ij
wn k
wn k
wn k
C kj n j C jk n j
E ik C ij n j n i C ij E jk
2C kjsym n j
(C kj C jk ) n j
C kj n j n i C ik
(1.348)
2C kj n j
C jk .
w C 1 C
wC
w1
wC
(1.349)
where O is the fourth-order zero tensor and the above equation in indicial notation
becomes:
w C iq1C qj
wC kl
wC C
1
iq
wC kl
w C iq1
wC kl
C iq1
qj
w C iq1
wC kl
whereas C qj
wC kl
wC kl
w C qj
wC kl
O ikjl
w C iq1
wC kl
1
qj C jr
C iq1
w C qj
wC kl
C jr1
(1.350)
C jr1
qr
C iq1
w C ir1
wC kl
C iq1
w C ir1
wC kl
w C qj
1
C qj C jq , so we can conclude that:
2
w C iq1
wC kl
w C qj
C iq1
qr
C iq1
qj
1 w C qj C jq 1
C jr
wC kl
2
>
1
E qk E jl E jk E ql C jr1
2
>
1 1 1
C ik C lr C il1 C kr1
2
>
1 1
C iq E qk E jl C jr1 C iq1E jk E ql C jr1
2
(1.351)
Or in tensorial notation:
wC 1
wC
>
1 1
C C 1 C 1 C 1
2
(1.352)
1 TENSORS
85
NOTE: Note that, if we had not replaced the symmetric part of C qj in (1.351), we would
w C iq1
wC kl
C iq1
qr
w C qj
wC kl
C jr1
C iq1E qk E jl C jr1
symmetric tensor.
1.5.11.1 Partial Derivative of Invariants
Let T be a second-order tensor. The partial derivative of I T with respect to T , (see
equation (1.342)), is:
w>I T @
wT
w>Tr ( T )@
wT
>Tr (T )@,T
(1.353)
>
>
w Tr ( T 2 )
1 w>Tr ( T )@
wT
wT
2
1
T
2( TrT )1 2 T
2
Tr ( T )1 T T
w 1
2
2
>Tr ( T )@ Tr ( T )
wT 2
>
(1.354)
1
III T T
I T 1 II T T
1
2
III T T
(1.355)
2
Tr ( T )1 T T
Tr ( T )1 I T 1 II T T 1 III T T 2
II
T
TT
1
III T T 2
(1.356)
To find the partial derivative of the third invariant, we can start with the definition given in
(1.313), so:
w> III T @
wT
3
1
1
w 1
3
2
Tr ( T ) Tr ( T ) Tr ( T ) >Tr ( T )@
2
6
wT 3
>
2
1
1 w Tr ( T 2 )
1
w>Tr ( T )@ 3
3( T 2 ) T
>Tr ( T )@ 1
Tr ( T ) Tr ( T 2 )
2
2
6
3
wT
wT
( T 2 ) T Tr ( T ) T T
2
1
1
Tr ( T 2 )1 >Tr ( T )@ 1
2
2
( T 2 ) T Tr ( T ) T T
2
1
>Tr(T )@ Tr( T 2 ) 1
2
>
(1.357)
( T 2 ) T I T T T II T 1
T I T T II T 1 III T T
III T T 1
1
T 2 I T T II T 1
(1.358)
86
III
TT
1 T
I T T II T 1
T
T
2 T
(1.359)
I T T T II T 1
By comparing (1.357) with (1.359) we find another way to express the derivative of III T
with respect to T , i.e.:
w> III T @
wT
III
TT
1 T
III T T T
(1.360)
T
D2
T
Dt 2
T
(1.361)
DTij
Dt
(1.362)
cof Tij
where cof Tij is the cofactor of Tij and defined as >cof Tij @T
detT T 1 ij .
w III b 2
wJ
wb
wb
1 w III
1
1
1
b
III b 2
III b 2 III b b T
2
2
wb
1
1
1
1
III b 2 b
J b 1
2
2
1
w ln III b 2
1 w III b 1 1
w>ln J @
b
wb
2 III b wb
2
wb
T sph T dev
Tr ( T )
1 T dev
3
IT
1 T dev
3
Tm 1 T dev
(1.363)
1 TENSORS
T dev
T
87
Tr ( T )
1 T Tm 1
3
(1.364)
Tijdev
T12dev
dev
T22
dev
T23
T13dev
dev
T23
T33dev
T12
T13
T11 Tm
T12
T
13
T12
T22 Tm
T23
T12
1
3
T33 Tm
T13
T23
T23
1
(2 T33 T11 T22 )
3
T13
(1.365)
Tr ( T )
Tr T
1
3
Tr ( T dev )
T dev
Tr ( T )
Tr ( T )
Tr (1) 0
3
E ii 3
(1.366)
Thus, we can conclude that the trace of any deviatoric tensor is equal to zero.
1.5.12.2 Second Invariant of the Deviatoric Tensor
For simplicity we can use the principal space to obtain the second and third invariant of the
deviatoric tensor. In the principal space the components of T are given by:
Tij
T1
0
0
T2
0
0
0
T3
(1.367)
T1 T2 T3 , II T
T1 Tm
0
Tijdev
0
T2 Tm
0
T1 T2 T2 T3 T3 T1 , III T
T3 Tm
0
0
T1 T2 T3 .
(1.368)
( T1 Tm )( T2 Tm ) ( T1 Tm )( T3 Tm ) ( T2 Tm )( T3 Tm )
( T1 T2 T1 T3 T2 T3 ) 2 Tm ( T1 T2 T3 ) 3Tm2
2I
I2
II T T ( I T ) T
3
3
1
3 II T I T2
3
(1.369)
88
We could also have obtained the above result, by directly starting from the definition of the
second invariant of a tensor given in (1.269), i.e.:
II T dev
1
2
1
2
1
2
1
2
^(I T
dev
>
) 2 Tr ( T dev ) 2
^ Tr>(T T
^ Tr>(T
> Tr(T
@`
^ >
1
Tr ( T dev ) 2
2
2 Tm T 1 Tm2 1)
m 1)
@`
@`
@`
) 2 Tm Tr ( T ) Tm2 Tr (1)
(1.370)
IT
I2
1
2
I T T 3
Tr ( T ) 2
2
3
9
I T2
1
2
Tr ( T )
2
3
2
3
II T dev
x3
2 I T2
1
2 II T
2
3
1
3 II T I T2
3
(1.371)
T 33
T23
T13
T13
T23
T12
T22
T12
x2
T11
x
1
x3
x3
Tm
T33dev
T23
T13
Tm
Tm
x1
T13
x2
T23
T12
T11dev
x1
dev
T22
T12
x2
1 TENSORS
89
T dev
>
1
Tr ( T dev ) 2
2
>
1
Tr ( T dev T dev )
2
1
T dev
2
T dev
1 dev dev
Tij T ji
2
(1.372)
T dev
>
1
dev 2
dev 2
( T11dev ) 2 ( T22
) ( T33dev ) 2 2( T12dev ) 2 2( T13dev ) 2 2( T23
)
2
(1.373)
T dev
1 dev dev
Tij T ji
2
>
1
( T1dev ) 2 ( T2dev ) 2 ( T3dev ) 2
2
(1.374)
dev
T22
dev
T23
dev
T23
T33dev
T11dev
T13dev
T13dev
T33dev
T11dev
T12dev
T12dev
dev
T22
>
1
dev dev
dev
dev 2
) ( T13dev ) 2
2 T22
T33 2 T11dev T33dev 2 T11dev T22
( T12dev ) 2 ( T23
2
(1.375)
or
2 T T T T 2 T T T
2 T T T T T T
1
dev
T22
2
II T dev
dev 2
11
dev
11
dev
22
dev
22
dev 2
33
dev
33
dev 2
22
dev 2
11
dev 2
11
dev
11
dev 2
22
dev
33
dev 2
33
dev 2
33
(1.376)
dev 2
) ( T13dev ) 2
( T12dev ) 2 ( T23
2 II
T dev
dev 2
2( T12dev ) 2 2( T13dev ) 2 2( T23
)
(1.377)
T dev
>
1
dev
dev 2
dev 2
( T22
T33dev ) 2 ( T11dev T33dev ) 2 ( T11dev T22
) ( T12dev ) 2 ( T23
) ( T13dev ) 2
6
(1.378)
Moreover, if we consider the principal space we obtain:
II
T dev
>
1
( T2dev T3dev ) 2 ( T1dev T3dev ) 2 ( T1dev T2dev ) 2
6
(1.379)
( T1 Tm )( T2 Tm )( T3 Tm )
T1 T2 T3 Tm ( T1 T2 T1 T3 T2 T3 ) Tm2 ( T1 T2 T3 ) Tm3
I
I2
I3
III T T II T T I T T
3
9
27
I T II T 2 I T3
III T
3
27
1
3
2 I T 9 I T II T 27 III T
27
(1.380)
90
T dev
(1.381)
ws
w
sph dev
sph s
I
1s
3
s
I
1.
3
Afterwards we calculate:
ws
w
Therefore
s ij
ws ij
wV kl
w 1
3
ws ij
wV ij
wV kl
wV kl
w> @ 1 w>I @
1
w
3 w
1 w >I @
E ij
3 wV kl
s ij E ik E jl E kl E ij
3
1
3
E ik E jl E kl E ij
s ij E ik E jl
1
s ij E kl E ij
3
1
s kl E kl s ii
,
3
0
s kl
s:
ws
w
To show that two tensors are coaxial, we must prove that dev dev :
dev
( sph ) sph
I
1
3
I
I
1 1
3
3
I
1 dev
3
Therefore, we have shown that and dev are coaxial tensors. In other words, they have
1 TENSORS
91
: ( T ) det T
: (T , S) T : S
(1.382)
where T and S are second-order tensors. Additionally, as an example of a second-ordervalued tensor function we have:
3
3 T B1 CT
(1.383)
1.6.1
The
function
f
f (x)
can
be
approximated
by
the
Taylor
series
as
1 w f (a)
( x a ) n , where n! denotes the factorial of n , and f (a) is the value of
wx n
n 0
the function at the application point x a . We can extrapolate that definition for use on
f ( x)
n!
tensors. For example, let us suppose we have a scalar-valued tensor function Z in terms of
a second-order tensor, E , then we can approximate Z (E ) as:
Z( E ) |
1
1 wZ ( E 0 )
1 w 2 Z( E 0 )
( E ij E 0 ij )
( E ij E ij 0 )( E kl E kl 0 )
Z( E 0 )
0!
1! wE ij
2! wE ij wE kl
wZ ( E 0 )
w 2 Z( E 0 )
1
| Z0
: (E E0 ) (E E0 ) :
: (E E0 )
wE
wE wE
2
(1.384)
S( E ) |
(1.385)
1S
(1.386)
92
exp S
1 O
a 1
ln(1 S )
O2a O3a
n ( a ) n ( a )
2!
3!
exp O
n ( a ) n ( a )
(1.387)
1 2 1 3
(a)
(a)
O a O a O a n n
2
3
a 1
ln(1 O
a)n
(a)
(a)
a 1
where O a and n ( a ) are the eigenvalues and eigenvectors, respectively, of the tensor S .
1.6.2
3 * T Q 3 T Q T
3 Q T QT
(1.388)
3 T*
We can show that 3 (T ) has the same principal directions of T , i.e. 3 (T ) and T are
coaxial tensors. To demonstrate this we can regard the components of T in the principal
space as:
T ij
O 1
0
0
O2
0
0
0
O 3
(1.389)
Q T QT
(1.390)
(1.391)
Qij
0
1 0
0 1 0
0 0 1
(1.392)
After having done the calculation for the matrices (1.391), we obtain:
3 11 3 12 3 13
3
3
3 22 3 23
12
3 13 3 23 3 33
0
0
3 11
0
3 * 0 3 22
0
0
3 33
*
3 11
3
12
3 13
3 12
3 22
3 23
3 13
3 23
3 33
3 13
3 23
(1.393)
0 . Therefore, 3 (T )
Once again we observe, a tensor function 3 ( T ) . This tensor function is isotropic if and
only if it can be represented by the following linear transformation, Truesdell & Noll
(1965):
1 TENSORS
3(T) ) 0 1 )1 T ) 2 T 2
93
(1.394)
O
Z
( a ) n ( a )
(1.395)
n ( a ) n ( a )
(1.396)
an
a 1
3
a 1
Note that T and 3 have the same principal directions n (i ) . Then, we can put the
following set of equations together:
1 n (1) n (1) n ( 2) n ( 2) n (3) n (3)
(1)
(1)
(2)
(2)
( 3)
( 3)
T O 1n n O 2 n n O 3n n
2
O21n (1) n (1) O23n ( 2 ) n ( 2 ) O23n (3) n (3)
T
(1.397)
O 2O3
O 2 O 3
T2
1
T
(O 1 O 3 )(O 1 O 2 )
(O 1 O 3 )(O 1 O 2 )
(O 1 O 3 )(O1 O 2 )
O 1 O 3
O 1O 3
T2
1
T
(O 2 O 1 )(O 2 O 3 )
(O 2 O 1 )(O 2 O 3 )
(O 2 O 1 )(O 2 O 3 )
(1.398)
O 1 O 2
O 1O 2
T2
1
T
(O 3 O 1 )(O 3 O 2 )
(O 3 O 1 )(O 3 O 2 )
(O 3 O 1 )(O 3 O 2 )
3(T) ) 0 1 )1T ) 2 T 2
(1.399)
Z1 O 2 O 3
Z 2 O 1O 3
Z 3 O 1O 2
(O 1 O 3 )(O 1 O 2 ) (O 2 O 1 )(O 2 O 3 ) (O 3 O 1 )(O 3 O 2 )
)1
)2
Z3
Z2
Z1
(O 1 O 3 )(O 1 O 2 ) (O 2 O 1 )(O 2 O 3 ) (O 3 O 1 )(O 3 O 2 )
Z1 O 2 O 3
Z 2 O 1 O 3
Z3 O 1 O 2
(O 1 O 3 )(O 1 O 2 ) (O 2 O 1 )(O 2 O 3 ) (O 3 O 1 )(O 3 O 2 )
(1.400)
We can now show that if a tensor function 3 ( T ) is given in (1.399), this tensor function is
isotropic:
94
3 * T Q 3 T Q T
Q ) 0 1 )1 T ) 2 T 2 Q T
) 0 Q 1 Q ) 1Q T Q ) 2 Q T
T
QT
) 0 1 )1T * ) 2 T *
(1.401)
3(T )
1.6.3
3 (A )
(1.402)
w3
(e i e j )
wA ij
(1.403)
3, AA
w 23
(e i e j e k e l ) D ijkl (e i e j e k e l )
wA ij wA kl
(1.404)
FT F
F FT
(1.405)
where F is an arbitrary second-order tensor with the restriction det ( F ) ! 0 imposed on it.
We must also bear in mind that there is a scalar-valued isotropic tensor function,
: : I C , II C , III C , expressed in terms of the principal invariants of C , where I C I b ,
II C II b , III C III b . Next, we can find the partial derivative of : with respect to C , and
with respect to b . We must also verify that the following relation holds:
F : ,C F T
: , b b
(1.406)
w: I C , II C , III C
wC
w: wI C
w: w II C
w: w III C
wI C wC w II C wC w III C wC
(1.407)
Considering the partial derivatives of the principal invariants, we can state that:
wI C
wC
w II C
wC
IC 1 C T
IC 1 C
w III C
wC
III C C T
III C C 1
II C C 1 III C C 2
C 2 I C C II C 1
(1.408)
wI C
wC
1,
w II C
wC
I C 1 C and
w:
w:
I C 1 C w: III C C 1
1
wI C
w II C
w III C
w III C
wC
III C C 1
(1.409)
1 TENSORS
: ,C
w:
w:
w:
I C 1
I
I
I
w
w
C
w II C
95
w:
C
III C C 1
I
I
I
w
C
w III C
wC
(1.410)
wI C
wC
1,
w II C
wC
IC 1 C
w:
w:
w:
w:
w:
IC
II C 1
I C C
w
w
w
w
I
I
I
I
I
I
w
I
I
I
I
I
C
C
C
C
w III C
: ,C
: ,C
w:
wI C
wI C
wC
1,
w II C
wC
II C C 1 III C C 2 ,
2
C
w III C
wC
(1.411)
III C C 1 , in the
w:
w:
w:
1
II C
III C C 1
III C C 2
w III C
w II C
w II C
I b , II C
: , b
wI b
II b , III C
(1.412)
w:
w:
w:
I b 1
III b b 1
b
w II b
w II b
w III b
(1.413)
w:
w:
w:
I C F 1 F T
F C F T
III C F C 1 F T
w
w
w
I
I
I
I
I
III C
w
C
C
C
F : , C F T
(1.414)
FT F
(1.415)
F C FT
C 1
F FT
F FT F FT
b b b2
F 1 b 1 F
F C 1 F T
F F 1 b 1 F F T
b 1 b
(1.416)
w:
w:
w: 2
w:
I C b
III C b 1 b
b
w
w
w
w
I
I
I
I
I
I
I
I
C
C
C
C
w:
w:
w:
w:
I C 1
III C b 1 b
b
w
w
w
w
I
I
I
I
I
I
I
I
C
C
C
(1.417)
In light of the equation in (1.413) and (1.417), we can draw the conclusion that:
F : , C F T
w:
w:
w:
w:
b
I b 1
III b b 1 b
w II b
w III b
wI b w II b
: ,b b b : ,b
(1.418)
96
:,F
w: C
wF
w: wC
:
wC wF
notation
indicial
o
: , F kl
w: wC ij
wC ij wFkl
(1.419)
wC ij
w Fqi Fqj
wFkl
wFkl
w Fqi
wFkl
Fqj Fqi
w Fqj
(1.420)
wFkl
E qk E il Fqj E qk E jl Fqi
E il Fkj E jl Fki
Then, by substituting (1.420) into (1.419), we obtain:
: , F kl
w:
E il Fkj E jl Fki
wC ij
Fkj
: , F kl
w:
Fkj
wC lj
w:
Fkj
wC jl
(1.421)
w:
w:
Fki
wC lj
wC il
:,F
2: , C F T
2F : , C
(1.422)
(1.423)
Therefore, we can draw the conclusion that : , C and U are coaxial tensors.
Let A be a symmetric second-order tensor, and :
function. The following relationships hold:
: (A ) be a scalar-valued tensor
: , b 2b : , A for A b T b
: , b 2: , A b for A b b T
2b : , A 2: , A b
: ,b
b : , A : , A b
for A b b and b b T
(1.424)
1 TENSORS
97
Tij
T11
T12
T13
T12
T22
T23
T13
T23 Voigt
o^T `
T33
T11
T
22
T33
T12
T23
T13
(1.425)
This representation is called the Voigt Notation. It is also possible to represent a secondorder tensor as:
E ij
E11
E12
E13
E12
E 22
E 23
E13
E 23 Voigt
o^E `
E 33
E11
E
22
E 33
2E12
2E 23
2E13
(1.426)
As we have seen before, a fourth-order tensor, C , that presents minor symmetry, i.e.
C ijkl C jikl C ijlk C jilk , has 6 u 6 36 independent components. Note that, due to the
symmetry of (ij ) we have 6 independent components, and due to the symmetry of (kl )
we have 6 independent components. In Voigt Notation we can represent these
components in a 6 -by- 6 matrix as:
>C @
C 1111
C
2211
C 3311
C 1211
C 2311
C 1311
C1122
C1133
C1112
C1123
C 2222
C 3322
C 2233
C 3333
C 2212
C 3312
C 2223
C 3323
C1222
C 2322
C1233
C 2333
C 1212
C 2312
C1223
C 2323
C1322
C1333
C 1312
C1323
C1113
C 2213
C 3313
C1213
C 2313
C1313
(1.427)
In addition to minor symmetry the tensor also has major symmetry, i.e. C ijkl C klij , and the
number of independent components have reduced to 21 . One can easily memorize the
order of the components in the matrix >C @ if we consider the order of the second-order
tensor in Voigt Notation, i.e.:
(11)
(22)
(33)
(13)
1.7.1
(1.428)
98
1 0 0
o^`
E ij { 1 0 1 0 Voigt
0 0 1
1
1
1
0
0
0
(1.429)
In the subsection 1.5.2.5.1 Unit Tensors we have defined three fourth-order unit tensors,
namely, I ijk"
E ik E j" , I ijk"
E ij E k" is a
symmetric tensor. The representation of I ijk" E ij E k" in Voigt notation can be evaluated by
observing how a symmetric fourth-order tensor is represented in (1.427), thus:
>@
E ij E k" Voigt
o I
I ijk"
1
E ik E j" E i" E jk , which in Voigt notation becomes:
2
o>I @
I ijk" Voigt
I1111
I
2211
I 3311
I1211
I 2311
I1311
(1.430)
components
1 1 0 0 0
1 1 0 0 0
1 1 0 0 0
0 0 0 0 0
0 0 0 0 0
0 0 0 0 0
The
of
1
1
0
0
fourth-order
unit
I1122
I 2222
I1133
I 2233
I1112
I 2212
I1123
I 2223
I 3322
I1222
I 3333
I1233
I 3312
I1212
I 3323
I1223
I 2322
I1322
I 2333
I1333
I 2312
I1312
I 2323
I1323
tensor,
I1113
I 2213
I 3313
I1213
I 2313
I1313
I sym ,
1
0
0
0
are
represented
0 0 0 0 0
1 0 0 0 0
0 1 0 0 0
0 0 12 0 0
0 0 0 12 0
0 0 0 0 12
by
(1.431)
>I @1
1.7.2
1
0
0
0
0 0 0 0 0
1 0 0 0 0
0 1 0 0 0
0 0 2 0 0
0 0 0 2 0
0 0 0 0 2
(1.432)
The dot product between a symmetric second-order tensor, T , and a vector n , is given by
&
&
&
b T n where the components of b can be evaluated as follows:
1 TENSORS
b1
b
2
b 3
T11
T
12
T13
T12
T22
T23
99
T13 n1 b1
T23 n 2 b 2
T33 n 3 b 3
(1.433)
b1
b
2
b 3
1.7.3
T11
T
22
n1 0 0 n 2 0 n 3
T
33
0 n
0 n1 n 3 0
2
T
0 0 n 3 0 n 2 n1 12
T23
>N @T
T13
^b`
>N @ ^T `
T
(1.434)
Tkl a ik a jl
(1.435)
or in matrix form:
T11c
T c
12
T13c
T12c
c
T22
c
T23
T13c
c
T23
c
T33
a11
a
21
a 31
a12
a 22
a 32
a13 T11
a 23 T12
a 33 T13
T12
T22
T23
T13 a11
T23 a 21
T33 a 31
a12
a 22
a 32
a13
a 23
a 33
(1.436)
By multiplying the matrices and by rearranging the result in Voigt notation we obtain:
^T c` >M@ ^T `
(1.437)
where:
^T c`
T11c
Tc
22
T33
c
;
T12c
T23
c
T13c
^T `
T11
T
22
T33
T12
T23
T13
(1.438)
and >M@ is the transformation matrix for the second-order tensor components in Voigt
Notation. The matrix >M@ is given by:
>M@
a11 2
2
a 21
a 2
31
a 21 a11
a a
31 21
a 31 a11
a12
a 22 2
a13
a 23 2
a 32 2
a 22 a12
a 33 2
a13 a 23
a 32 a 22
a 32 a12
a 33 a 23
a 33 a13
2a11 a12
2a 21 a 22
2a12 a13
2a 22 a 23
2a 31 a 32
2a 32 a 33
2a 31 a 33
(1.439)
a13 a 21 a11 a 23
a 33 a 21 a 31 a 23
a 33 a11 a 31 a13
2a11 a13
2a 21 a 23
100
^E c` >N @^E `
(1.440)
where
>N @
a11 2
2
a 21
2
a
31
2a 21 a11
2a a
31 21
2a 31 a11
a12 2
a 22 2
a 32
a13 2
a 23 2
a11 a12
a 21 a 22
a12 a13
a 22 a 23
a 31 a 32
a 32 a 33
a 33
2a 22 a12
2a 32 a 22
2a13 a 23
2a 33 a 23
2a 32 a12
2a 33 a13
a 31 a 33
a13 a 21 a11 a 23
a 33 a 21 a 31 a 23
a 33 a11 a 31 a13
a11 a13
a 21 a 23
(1.441)
The matrices (1.439) and (1.441) are not orthogonal matrices, i.e. >M@ z >M@T and
>N @1 z >N @T . However, it is possible to show that >M@1 >N @T .
1
1.7.4
n ( a ) n ( a )
Matricial
form
o
a 1
AT T c A
(1.442)
where A is the transformation matrix between the original set and the principal space,
made up of the eigenvectors n ( a ) . The above equation can be rewritten in terms of
components as follows:
T11
T
12
T13
T12
T22
T23
T13
T1
T23 A T 0
0
T33
0 0
0 0
0 0 A A T 0 T2
0 0
0 0
0
0 0 0
0 A A T 0 0 0 A
0 0 T3
0
(1.443)
or
T11
T
12
T13
T12
T22
T23
T13
T23
T33
2
a11
T1 a11 a12
a a
11 13
2
a 31
T3 a 31 a 32
a a
31 33
a11 a12
2
a12
a12 a13
a 31 a 32
2
a 32
a 32 a 33
2
a 21
a11 a13
a12 a13 T2 a 21 a 22
2
a a
a13
21 23
a 21 a 22
2
a 22
a 22 a 23
a 31 a 33
a 32 a 33
2
a 33
a 21 a 23
a 22 a 23
2
a 23
(1.444)
By regarding how second-order tensors are presented in Voigt Notation as in (1.438), the
spectral representation of a second-order tensor in Voigt notation becomes:
^T `
T11
T
22
T33
T12
T23
T13
2
2
2
a 31
a 21
a11
2
2
2
a
a
a
32
22
12
a2
a2
a2
13
23
T1
T3 33
T2
a 31 a 32
a 21 a 22
a11 a12
a a
a a
a a
32 33
22 23
12 13
a 31 a 33
a 21 a 23
a11 a13
(1.445)
1 TENSORS
1.7.5
101
T12
T13
T12
1
3
T23
1
(2 T33 T11 T22 )
3
T13
(1.446)
22
T33dev 1 1 1 2 0 0 0 T33
dev
(1.447)
0
0 3 0 0 T12
T12 3 0
T dev
0
0
0 0 3 0 T23
23
dev
0
0 0 0 3 T13
0
T13
&
Problem 1.40: Let T ( x , t ) be a symmetric second-order tensor, which is expressed in
&
terms of the position ( x ) and time (t ) . Also, bear in mind that the tensor components,
x2
T22
T22
T12
T12
T12
T11
T11
T11
x1
T12
x3
T22
x1
102
x2
xc1
xc2
R
x1
2
a 21
a a
21 11
T11c
Tc
22
T12c
a12
a 22
a 22 a12
T11
2a 21 a 22
T22
a11 a 22 a12 a 21 T12
2a11 a12
(1.448)
T c A T AT
x2
Tc22
xc2
T22
Tc12
Tc11
T12
T11
T11
Tc11
T12
Tc12
x1
T22
xc1
P
T
Tc22
x1
AT T c A
a 32
a 33
Making
use
of
cos 2 R
sin 2 R
2 cos R sin R T11
2
2
cos R
2 sin R cos R T22
sin R
sin R cos R cos R sin R cos 2 R sin 2 R T
12
the
following
trigonometric
1 cos 2R
, cos 2 R
2
(1.450)
1 TENSORS
T11c
Tc
22
T12c
1 cos 2R
1 cos 2R
R
2
sin
103
1 cos 2R
sin 2R
2
T11
1 cos 2R
T
2
R
sin
22
2
T12
sin 2R
cos 2R
1 cos 2R
1 cos 2R
T11
T22 T12 sin 2R
T11c
2
2
1 cos 2R
1 cos 2R
c
T11
T22 T12 sin 2R
T22
2
2
sin 2R
sin 2R
T12c
T22 T12 cos 2R
T11
2
2
T11 T22
T12c
sin 2R T12 cos 2R
(1.451)
b) Recalling that the principal directions are characterized by the lack of any tangential
components, i.e. Tij 0 if i z j , in order to find the principal directions in the plane, we let
T12c 0 , hence:
T T22
T T22
11
sin 2R T12 cos 2R 0 11
sin 2R
2
2
2 T12
2 T12
sin 2R
tg(2R )
cos 2R T11 T22
T11 T22
T12c
T12 cos 2R
(1.452)
To find the principal values (eigenvalues) we must solve the following characteristic
equation:
T11 T
T12
T12
T22 T
T11 T22
r
2
By rearranging the above equation we obtain the principal values for the two-dimensional
case as:
104
T11 T22
T T22
r 11
2
2
T(1, 2 )
(1.453)
T122
c) We directly apply equation (1.451) to evaluate the values of the components Tijc ,
(i, j 1,2) , where T11 1 , T22 2 , T12 4 and R 45 , i.e.:
1 2 1 2
T11c 2 2 cos 90 4 sin 90 2.5
1 2 1 2
c
cos 90 4 sin 90 5.5
T22
2 2
1 2
T12c
sin 90 4 cos 90 0.5
T 41.4375
arctg
2
1 2
T11 T22
&
The principal values of T ( x , t ) can be evaluated as follows:
T11 T22
T T22
r 11
2
2
T(1, 2 )
T122
T1
T2
5.5311
2.5311
T2
41.437
x1c
Components
V2
T1
131.437
c
T22
5.5311
T22
T12c
T11
0
0
50
100
-2
45
T12
-4
T2
150
200
250
300
T11c
x1c
2.5311
T
86.437
-6
TS max
4.0311
350
1 TENSORS
105
&
A tensor field indicates how the tensor, T ( x , t ) , varies in space ( x ) and time ( t ). In this
section, we regard the tensor field as a differentiable function of position and time. For
more information about it, we need to define some operators, e.g. gradient, divergence, curl,
which we can use as indicators of how these fields vary in space.
A tensor field which is independent of time is called a stationary or steady-state tensor
&
field, i.e. T T ( x ) . However, if the field is only dependent on t then it is said to be
&
homogeneous or uniform. That is, T (t ) has the same value at every x position.
Tensor fields can be classified according to their order as: scalar, vector, second-order
&
tensor fields, etc. As an example of a scalar field we can quote temperature T ( x , t ) and in
Figure 1.34(a) we can see temperature distribution over time t t1 . Then, as an example of
& &
a vector field we can quote velocity v ( x , t ) and Figure 1.34(b) shows velocity distribution,
&
in which each point is associated with a vector v over time t t1 .
x3
t
T5
T8
x3
t1
t1
& &
v ( x , t1 )
&
T4 ( x ( 4) , t1 )
T6
T3
T7
T1
x2
x2
T2
x1
x1
a) Scalar field
b) Vector field
G G( x, t )
(1.454)
&
v
vi
& &
v ( x, t )
&
v i ( x, t )
(1.455)
T
Tij
&
T ( x, t )
&
Tij ( x , t )
(1.456)
Vector Field
Tensorial notation
Indicial notation
Second-Order Tensor Field
Tensorial notation
Indicial notation
106
1.8.1
Scalar Fields
&
The next analysis is carry out with reference to a stationary scalar field, i.e. G G( x ) , with
continuous values of wG / wx1 , wG / wx 2 and wG / wx 3 . Then, observe that the value of the
&
&
&
&
scalar function at point ( x ) is G ( x ) , and if we observe a second point located at ( x dx ) ,
the total derivative (differential) of the function G is defined as:
&
&
&
G ( x dx ) G ( x ) { dG
G ( x1 dx1 , x 2 dx 2 , x3 dx3 ) G ( x1 , x 2 , x3 ) { dG
(1.457)
dG
(1.458)
dG G, i dxi
1.8.2
(1.459)
Gradient
(1.460)
where the operator x& is known as the Nabla symbol. Expressing the equation (1.460) in
the Cartesian basis we obtain:
wG
wG
wG
dx1
dx 2
dx3
wx3
wx 2
wx1
>( G) x e
&
x
( x&
G) x2 e 2 ( x& G) x3 e 3
@ >(dx )e
1
(dx 2 )e 2 (dx3 )e 3
(1.461)
(1.462)
Therefore, we can draw the conclusion that the x& G components in the Cartesian basis
are:
( x& G )1 {
wG
wx1
( x& G) 2 {
wG
wx 2
( x& G) 3 {
wG
wx3
(1.463)
wG
wG
wG
e1
e2
e3
wx3
wx 2
wx1
(1.464)
1 TENSORS
107
w
e i { w , i e i Nabla symbol
wxi
x&
(1.465)
The direction of
The magnitude of
x& G
x& G
(1.466)
The surface G const , called the surface level, or isosurface or equiscalar surface, is the
surface formed by points which all have the same value of G , so, if we move along the
level surface the values of the function do not change.
& &
(1.467)
Using the definition of x& , given in (1.465), the gradient of the vector field becomes:
&
x& v
w ( v i e i )
e j
wx j
( v i e i ) , j e j
v i , j e i e j
(1.468)
x& G
G const c1
G const c 2
c1 ! c 2 ! c3
G const c 3
Therefore, we can define the gradient of a tensor field (x( x , t )) in the Cartesian basis as:
x& (x)
(1.469)
As noted, the gradient of a vector field becomes a second-order tensor field, whose
components are:
108
v i, j {
wv i
wx j
wv 1
wx1
wv 2
wx
1
wv 3
wx1
wv 1
wx3
wv 2
wx3
wv 3
wx3
wv 1
wx 2
wv 2
wx 2
wv 3
wx 2
(1.470)
&
w (Tij e i e j )
wx k
e k
Tij ,k e i e j e k
(1.471)
x& T ijk
{ Tij ,k
(1.472)
Problem 1.41: Find the gradient of the function f ( x1 , x 2 ) cos( x1 ) exp x1x2 at the point
( x1 0, x 2 1) .
Solution: By definition, the gradient of a scalar function is given by:
x& f
where:
wf
wx1
x& f ( x1 , x 2 )
wf
wf
e1
e2
wx1
wx 2
wf
;
x1 exp x1x2
wx 2
x1x2
x1x2
x& f (0,1)
>1@
e 1 >0@ e 2
& &
2e 1
Problem 1.42: Let u( x ) be a stationary vector field. a) Obtain the components of the
&
& &
differential du . b) Now, consider that u( x ) represents a displacement field, and is
independent of x3 . With these conditions, graphically illustrate the displacement field in
the differential area element dx1 dx 2 .
Solution: According to the differential and gradient definitions, it holds that:
& &
u( x )
x2
&
x
&
dx
& &
&
u( x dx )
&
&
x dx
x1
x3
du i
wu i
dx j
wx j
du1
du
2
du 3
wu1
wx1
wu 2
wx
1
wu 3
wx1
wu1
wx 2
wu 2
wx 2
wu 3
wx 2
wu1
wx3 dx
1
wu 2
dx
2
wx3
wu 3 dx3
wx3
1 TENSORS
109
or:
du1
du 2
du 3
wu1
wu
wu
dx1 1 dx 2 1 dx3
wx1
wx 2
wx3
wu 2
wu
wu
dx1 2 dx 2 2 dx3
wx1
wx 2
wx3
wu 3
wu
wu
dx1 3 dx 2 3 dx 3
wx1
wx 2
wx3
with
du1 u1 ( x1 dx1 , x 2 dx 2 , x3 dx3 ) u1 ( x1 , x 2 , x3 )
du 2 u 2 ( x1 dx1 , x 2 dx 2 , x3 dx3 ) u 2 ( x1 , x 2 , x3 )
du u ( x dx , x dx , x dx ) u ( x , x , x )
3
1
1
2
2
3
3
3
1
2
3
3
As the field is independent of x3 , the displacement field in the differential area element is
defined as:
wu1
wu1
1
2
w
w
u
u
2
du u ( x dx , x dx ) u ( x , x )
dx1 2 dx 2
2
1
1
2
2
2
1
2
2
wx1
wx 2
or:
wu1
wu1
u1 ( x1 dx1 , x 2 dx 2 ) u1 ( x1 , x 2 ) wx dx1 wx dx 2
1
2
w
w
u
u
2
u ( x dx , x dx ) u ( x , x )
dx1 2 dx 2
1
2
2
2
1
2
2 1
wx1
wx 2
Note that the above equation is equivalent to the Taylor series expansion taking into
account only up to linear terms. The representation of the displacement field in the
differential area element is shown in Figure 1.36.
110
u2
wu 2
dx 2
wx 2
u2
( x1 dx1 , x 2 dx 2 )
( x1 , x 2 dx 2 )
u1
wu
wu 2
dx1 2 dx 2
wx 2
wx1
wu1
dx 2
wx 2
u1
&
du
dx 2
wu1
wu
dx1 1 dx 2
wx 2
wx1
u2
(u 2 )
( x1 dx1 , x 2 )
( x1 , x 2 )
x2
wu 2
dx1
wx1
u1
(u1 )
wu1
dx1
wx1
dx1
x1
=
x 2 ,u 2
u2
wu1
dx2
wx2
wu 2
dx2
wx2
Bc
dx 2
Ac
Oc
u2
Bc
dx 2
Ac
Oc
dx1
u1
u1
wu 2
dx1
wx1
dx1
wu1
dx1
wx1
x1 ,u1
1 TENSORS
1.8.3
111
Divergence
& &
(1.473)
&
&
&
div ( v ) { x& v x& v : 1
(1.474)
Then:
&
&
x& v x& v : 1
>v
@>
i , j e i e j : E kl e k e l
wv 1 wv 2 wv 3
wx1 wx 2 wx3
v i , j E kl E ik E jl
v k ,k
(1.475)
or
&
x& v
&
x& v : 1
>v e e @: >E
>v E E e @ e
>v e @ e
w>v e @
e
i, j
i, j
kl
i ,k
lj
e l
kl e k
(1.476)
wx k
Which we can use to insert the following operator into the Cartesian basis:
x& (x)
(1.477)
We can also verify that, when divergence is applied to a tensor field its rank decreases by
one order.
&
w(Tij e i e j )
wTij
wx k
E jk e i
e k
(1.478)
wx k
Tik ,k e i
&
NOTE: In this text book, when dealing with gradient or divergence of a tensor field, e.g. x& v
(the gradient of the vector field), x& T (the gradient of a second-order tensor field), x& T
(divergence of a second-order tensor field), this does not indicate that we are making a
&
&
&
&
tensor operation between a vector and a tensor, i.e. x& v z ( x& ) ( v ) , x& T z ( x& ) ( T )
&
and x& T z ( x& ) ( T ) and so on. In this textbook, x& is an operator which must be
applied to the entire tensor field, so, the tensor must be inside the operator, (see equations
&
(1.477) and (1.469)). Nevertheless, it is possible to relate x& v , x& T or x& T to tensor
operations between tensors, and it is easy to show that:
112
&
&
( v ) ( x& )
&
x& T ( T ) ( x& )
&
&
x& T ( T ) ( x& ) ( x& ) ( T T )
&
x& v
(1.479)
Once the Nabla symbol is defined we introduce the Laplacian operator 2 as:
x&
x&
wxi
x& x&
e e
wx j j i
w2
w2
w2
{ 2 2 2
wx1 wx 2 wx3
w, k w , k
w w
E ij
wxi wx j
w2
wxi wxi
(1.480)
w, kk
& &
>
2&
x& v
&
2&
x& ( x& v ) components
o x& v
&
&
and b
&
&
&
&
x& (a b) x& a x& b holds.
wx i
e i
wa j
wx i
e j e i
wb k
e k e i
wx i
v i ,kk
(1.481)
& &
w
, we can express x& (a b) as:
e i
wx i
&
&
(a i b i ), i
a i , i b i , i
&
wa i wb i
wx i wx i
&
&
x& a x& b
&
&
x& a x& b
&
&
w
( i 1,2,3 ), the following is
e i
wx i
true:
wa j
& & w (a j e j )
wa j
e i (b k e k )
( x& a) b
e j e i (b k e k ) b k E ik
e j
wx i
w
wx i
x
i
Thus,
j 1
b1
wa1
wa
wa
b2 1 b3 1
wx1
wx 2
wx 3
2 b1
wa 2
wa
wa
b2 2 b3 2
wx1
wx 2
wx 3
3 b1
wa 3
wa
wa
b 2 3 b3 3
wx1
wx 2
wx 3
bk
wa j
wx k
e j
1 TENSORS
113
&
q
x&
T
1 & &
1 &
x q 2 q x& T
T
T
& &
&
where q( x , t ) is an arbitrary vector field, and T ( x , t ) is a scalar field.
Solution:
&
q
T
&
x
1.8.4
w qi qi
1
1
q i ,i 2 q i T,i
{
wx i T T ,i T
T
1 & &
1 &
x q 2 q x& T (scalar)
T
T
The Curl
&
&
&
&
The curl (or rotor) of a vector field, v ( x ) is denoted by curl( v ) { rot ( v ) { x& v , and is
defined in the Cartesian basis as:
&
x& (x)
w
e j (x) The curl (rotor) of a tensor field in
wx j
the Cartesian basis
(1.482)
Note that the curl is already a tensor operator between two vectors. Using the definition of
the vector product we obtain the curl of a vector field as:
&
&
&
rot ( v ) x& v
w
e j ( v k e k )
wx j
wv k
e j e k
wx j
wv k
. ijk e i
wx j
. ijk v k , j e i
(1.483)
where . ijk is the permutation symbol defined in (1.55). Moreover, we have applied the
definition e j e k
&
&
&
rot ( v ) x& v
e 1
w
wx1
v1
wv 3
wx 2
e 2
w
wx 2
v2
wv
2
wx3
e 3
w
. ijk v k , j e i
wx3
v3
wv
wv
wv
wv
e 1 1 3 e 2 2 1 e 3
wx1 wx 2
wx3 wx1
(1.484)
We can verify that the antisymmetric part of a vector field gradient, which is illustrated by
&
( x& v ) skew { W , has as components:
>(
&
skew
&
x v)
ij
{ v iskew
,j
0
W
21
W31
1 wv 2 wv 1
2 wx1 wx 2
1 wv
wv
3 1
2 wx1 wx3
W12
0
W32
W13
W23
0
1 wv 1 wv 2
2 wx 2 wx1
0
1 wv 3 wv 2
2 wx 2 wx3
0
W
12
W13
W12
0
W23
1 wv 1 wv 3
2 wx 3 wx1
1 wv 2 wv 3
2 wx 3 wx 2
W13
W23
0
0
w
3
w2
w3
0
w1
(1.485)
w2
w1
0
114
&
where w1 , w2 , w3 are the components of the axial vector w associated with W , (see
subsection: 1.5.2.2.2. Antisymmetric Tensor).
With reference to the definition of the curl in (1.484) and the relationship in (1.485), we can
conclude that:
wv
wv 3 wv 2
wv
wv
wv
e 1 1 3 e 2 2 1 e 3
wx1 wx 2
wx3 wx1
wx 2 wx3
2W32 e 1 2W13 e 2 2W21e 3
2w1 e 1 w2 e 2 w3 e 3
&
2w
&
&
&
rot ( v ) { x& v
(1.486)
& &
wv
&
1 && &
x v v
2
(1.487)
It could be interesting to note that the equation in (1.486) can be obtained by means of
Problem 1.18, in which we showed that
&
&
1 & &
(a x ) is the axial vector associated with the
2
antisymmetric tensor ( x a ) skew . Therefore, the axial vector associated with the
&
>(v& ) (& )@
skew
is the vector
1 && &
x v .
2
As we can see, the curl describes the rotational tendency of the vector field.
Summary
Divergence
x
div (x)
{ x&
Scalar
Gradient
grad(x) {
x&
Curl
&
rot (x) { x& x
vector
Vector
Scalar
Second-order tensor
Vector
Second-order tensor
Vector
Third-order tensor
Second-order tensor
&
&
&
&
&
&
rot (Oa) x& (Oa) O( x& a) ( x& O a)
&
&
The result of the algebraic operation x& (Oa) is a vector, whose components are
given by:
>&
&
x
&
(Oa) i
. ijk (Oa k ) , j
. ijk (O , j a k Oa k , j )
. ijk Oa k , j . ijk O , j a k
&
O( x& a) i . ijk ( x& O ) j a k
&
&
O( x& a) i ( x& O a) i
(1.488)
1 TENSORS
115
& &
& &
&
& &
& &
& &
x& (a b) ( x& b)a ( x& a)b ( x& a) b ( x& b) a
& &
& &
The components of the vector product (a b) are given by (a b) k
>&
&
x
& &
( a b) l
>&
&
x
& &
( a b) l
. lpk (. kij a i b j ) , p
(1.489)
. kij a i b j , thus:
(1.490)
. kij . lpk (a i , p b j a i b j , p )
. kij . lpk (a i , p b j a i b j , p )
(E il E jp E ip E jl )(a i , p b j a i b j , p )
(1.491)
E il E jp a i , p b j E ip E jl a i , p b j E il E jp a i b j , p E ip E jl a i b j , p
al , p b p a p, p b l al b p, p a p b l, p
>
&
&
>
>
& &
a l , p b p , ( x& a)b l
& &
( x& b) a l a p b l , p .
&
&
&
&
2&
>
& &
a p , p b l , ( x& b)a l
al b p, p ,
(1.492)
. ijk a k , j , thus:
ci
>
&
&
&
x& ( x& a) q
. qli c i ,l
>&
&
x
&
&
( x& a) q
. qli (. ijk a k , j ) ,l
. qli . jki
E qj E lk a k , jl E qk E lj a k , jl
a k ,kq a q ,ll
>
2&
a k , kq and x& a q
(1.494)
a q ,ll .
(1.493)
(E qj E lk E qk E lj )a k , jl
. qli . ijk a k , jl
. qli . ijk a k , jl
GZ ,ii G ,i Z ,i
(1.495)
G x& 2 Z ( x& G ) ( x& Z )
(1.496)
where G and Z are scalar fields. Other interesting equations derived from the above are:
x& (G x& Z ) G x& Z ( x& G ) ( x& Z )
2
(1.497)
1.8.5
(1.498)
(1.499)
116
& &
&
&
&
&
other words, given a conservative field, the curl x& b equals zero. However, if the curl
of a vector field equals zero, this does not necessarily mean that the field is conservative.
&
Problem 1.46: Let G be a scalar field, and u be a vector field. a) Show that
&
&
x& v
&
x&
0 and x& x& G
&
&
& &
b) Show that x& x& v v
&
&
c) Referring x& v , show that
>
&
0.
>
&
&
& &
&
& &
&
&
( x& v )( x& v ) x& ( x& v ) v ( x& v ) ( x& v ) ;
&
&
&
2&
2
2 &
x& ( x& v ) x& ( x& v ) x& .
Solution:
&
&
Regarding that: x& v . ijk v k , j e i
x&
&
w
w
w
. ijk v k , j e i e l . ijk
v k , j E il . ijk
v k , j . ijk v k , ji
wx l
wx l
wx i
&
The second derivative of v is symmetrical with ij , i.e. v k , ji v k ,ij , while . ijk is
x&
&
v
. jik , thus:
. ijk v k , ji
. ij1v1, ji . ij 2 v 2, ji . ij 3 v3, ji
We can observe that . ij1v1, ji equals the double scalar product by using a symmetric and an
antisymmetric tensor, so . ij1v1, ji 0 .
Likewise, we can show that:
&
&
x& x& G . ijk G , kj e i 0 i e i 0
&
&
&
>
&
&
& &
x& x& v v
&
& &
x& ( v )
&
& &
& &
& &
& &
& &
x& ( v ) ( x& v ) ( x& )v ( x& ) v ( x& v )
&
&
&
Note that x& x& ( x& v ) 0 . Then, we can draw the conclusion that:
&
& &
& &
& &
& &
x& ( v ) ( x& v ) ( x& ) v ( x& v )
&
&
& &
&
& &
&
&
( x& v )( x& v ) x& ( x& v ) v ( x& v ) ( x& v )
>
&
&
&
&
2&
x& v x& ( x& v ) x& ( x& v )
&
&
&
x& ( x& v ) x&
&
&
&
&
&
&
2&
x& ( x& v ) x& > x& ( x& v )@ x& ( x& )
&
0
&
&
&
Referring once again to the equation in (1.492) to express the term x& ( x& ) :
&
2&
x& ( x& v )
&
&
&
x& ( x& )
&
&
2
x& ( x& v )
&
2&
x& ( x& ) x&
>
&
&
2&
x& x& ( x& v ) x&
0
1 TENSORS
117
Integration by Parts
u( x)vc( x)dx
u ( x )v ( x )
where v c( x)
1.9.2
v( x)u c( x)dx
(1.500)
dv
, and the functions u (x) , v(x) are differentiable in a d x d b .
dx
Given a domain B with a volume V , and bounded by the surface S , (see Figure 1.37), the
divergence theorem, also called the Gauss theorem, applied to the vector field states that:
&
x
&
&
i ,i
&
dV
&
v n dS v dS
dV
v n
i
dS
(1.501)
dS i
x2
&
dS
dS
B
&
x
x1
x3
Figure 1.37.
Let T be a second-order tensor field defined in the domain B . The divergence theorem
applied to this field is defined as:
&
x
dV
&
T n dS T dS
S
ij , j
dV
T n
ij
dS
ij
dS j
(1.502)
118
(x
(E
), j dV
ik
xi ), j dV
>E
E ik xi , j dV
ik , j x i
dV
k, j
x n
k
dS
x n
dS
(1.503)
E kj dV
x n
k
x i n j dS
ik
x n
VE kj
dS
dS
&
x n dS
V1
E kj , we
(1.504)
(x V
i
jk
(x V
), k dV
jk
), k dV
x V
i
jk n k
dS
>x
i , k V jk
>E
ik V jk
xi V jk ,k dV
x V
jk n k
dS
x V
jk n k
dS
(1.505)
xi V jk ,k dV
x V
i
jk , k
xV
dV
jk n k
&
x x& dV
dS V ji dV
V
&
x ( n ) dS T dV
(1.506)
or
&
&
x
&
&
x dS
dV
dV
(1.507)
&
x
&
) x dV
&
( n ) x dS dV
S
(1.508)
>m :
&
&
x ( x X )
1 TENSORS
>m
ij X , ij
@ d8 ( X ,
(
119
>m
m ij )n j d(
ij , j X , i
@ d8
x2
x1
Figure 1.38
Solution: We could directly apply the definition of integration by parts to demonstrate the
above relationship. But, here we will start with the definition of the divergence theorem.
&
That is, given a tensor field v , it is true that:
&
x
&
&
v n d( o 8 v
(
d8
&
indicial
d8
j, j
jn j
d(
&
Observing that the tensor v is the result of the algebraic operation v x& X m and the
equivalent in indicial notation to v j X, i m ij , and by substituting it in the above equation
we obtain:
j, j
d8
jn j
d(
>X,
m ij
>X,
ij
,j
X,
d8
>X,
ij
m ij n j d(
m ij X , i m ij , j d8
m ij d8
X,
X,
m ij n j d(
m ij n j d(
>X,
m ij , j d8
>m :
&
&
x ( x X )
NOTE: Consider now the domain defined by the volume V , which is bounded by the
&
surface S with the outward unit normal to the surface n . If N is a vector field and T is a
scalar field, it is also true that:
N T,
i
ij
dV
N T,
i
&
N x& ( x& T )dV
n j dS N i , j T , i dV
(
S
&
xT
&
&
N ) n dS x& T x& N dV
V
120
1.9.3
Independence of Path
A curve which connects two points A and B is denoted by the path from A to B , (see
Figure 1.39). We can then establish the condition by which a line integral is independent of
path, (see Figure 1.39).
C1
&
dr
&
b
If
& &
b dr
C1
x3
& &
b dr
C2
&
b - Conservative field
C2
x2
x1
& &
Let b( x ) be a continuous vector fields, then the integral
&
&
&
b dr
C
and only if b is a conservative field. This means that there is a scalar field G so that b x& G .
Regarding the above, we can draw the conclusion that:
B
& &
b dr
&
x
&
G dr
&
(b1 e 1 b 2 e 2 b 3 e 3 ) dr
wG
wG
wG &
e 1
e2
e 3 dr
x
x
x3
w
w
w
1
2
(1.509)
Thus
b1
wG
wx1
wG
wx 2
; b2
; b3
wG
wx3
(1.510)
&
e 1
w
wx1
b1
e 2
w
wx 2
b2
e 3
w
wx3
b3
0i
(1.511)
wx 2 wx3
wb1 wb 3
wx3 wx1
wb 2 wb1
wx1 wx 2
0
0
0
wb 3
wx 2
wb
1
wx3
wb 2
wx1
wb 2
wx 3
wb 3
wx1
wb1
wx 2
Therefore, if the above condition is not satisfied, the field is not conservative.
(1.512)
1 TENSORS
1.9.4
121
Let S be a regular surface, (see Figure 1.40), and F( x , t ) be a vector field. According to the
Kelvin-Stokes Theorem:
& &
&
&
&
&
&
(1.513)
F d( ( x& F) dS ( x& F) n dS
(
If p denotes the unit vector tangent to the boundary ( , the Stokes theorem becomes:
&
&
F p d( 8 (
(
&
x
&
&
F ) dS
&
(
8
&
x
&
F) n dS
(1.514)
&
&
&
x& F
e 1
w
wx1
F1
e 2
w
wx 2
F2
e 3
w
wx3
F3
wF3 wF2
wx 2 wx3
wF
wF
e 1 1 3
w
x
3 wx1
wF
wF
e 2 2 1
w
x
w
x2
1
e 3
(1.515)
dS 3
(1.516)
F dx
(
1
F2 dx 2 F3 dx3
wF3
wx
8
wF2
wx3
wF
wF
dS1 1 3
x
w
wx1
3
wF
wF
dS 2 2 1
x
x2
w
w
1
x3
(
x2
x1
which is known as the Stokes theorem in the plane or Greens theorem, which is expressed in
terms of components as:
F dx
(
1
F2 dx 2
wF2
wx
8
wF1
wx 2
dS 3
(1.518)
122
x3
(
8
x2
x1
Figure 1.41.
&
dS
x3
dSe 3
x2
e 3
x1
1.9.5
Greens Identities
&
&
x
&
&
F n dS
dV
(1.519)
(1.520)
(1.521)
&
and also regarding that F G x& Z , and by substituting (1.520) into (1.519) we obtain:
1 TENSORS
&
x
Z ( x& G) ( x& Z) dV
123
&
x
Z n dS
( x&
G)
( x&
Z) dV
x&
(1.522)
Z n dS G x& 2 Z dV
&
x
(G
Z Z x& 2 G dV
&
x
Z Z x& G) n dS
(1.523)
&
&
&
Problem 1.48: Let b be a vector field, which is defined as b x& v . Show that:
Ob n
i
O,
dS
&
b i dV
where O O( x ) .
&
&
&
Ob n
i
dS
O.
dS
ijk v k , j n i
Ob n
i
dS
O.
S
dS
ijk v k , j n i
(.
ijk O v k , j ), i
(.
ijk O , i
dV
v k , j . ijk Ov k , ji ) dV
(O,
. ijk v k , j O . ijk v k , ji ) dV
bi
O,
b i dV
Appendix
Graphical Representation
of a Second-Order Tensor
where T N
T n whose
&
&
&
t (n ) T N T S
&
is the normal vector, and TS is the tangential vector, (see Figure A.1).
(A.1)
&
If we then bear in mind that n and s are unit vectors according to the directions T N and
&
&
T S , respectively, the vector t (n) can also be expressed as:
&
t (n) T N n TS s
&
&
where T N and TS are the magnitudes of T N and TS , respectively.
&
The vector T N can also be evaluated as follows:
&
TN i
T N n i
TN
T N n { T N n
& (n )
& (n )
t (kn) n k ) n i (n k t (kn) ) n i
(
(t n) n (n t ) n
@n
> n
T
n
TN
>
n T n @n
k
kj
(A.2)
(A.3)
TN
125
126
Thus:
&
t (n) n n T n n k Tkj n j
TN
(A.4)
&
t (n)
x3
&
TN
&
TS
P
e 3
e 1
e 2
x2
x1
TS s
&
(t (n) s )s
&
TS i
&
(s t (n) )s
TS s i
(t (jn) s j )s i
> T
n s @ s
>
s
T
n @ s
jk
TS
(A.5)
i
TS
Then, another way to evaluate the tangential vector can be by means of the equation in
(A.1):
&
TS
&
&
t (n) T N
>
T n T : (n n ) n
(A.6)
&
Note, the magnitude of TS can also be obtained by means of the Pythagorean Theorem,
i.e.:
&
t (n)
&
TN
&
TS
TS2
t i(n) t (i n) T N2
(A.7)
Tij Tik n j n k .
The question now is: on what plane is the maximum normal and tangential component?
The answer to this problem is related to the maximum and minimum values of a function,
which will be discussed in the next subsection.
127
&
&
&
t (n) (1 n n ) , where t (n) is vector resulting from
&
projecting the second-order tensor T onto the n -direction, and T S is the tangential
vector.
&
Solution 1: If we consider that t (n)
&
TS
&
t (n)
&
&
T N TS and (A.3) we can state that:
&
&
&
&
t (n) n n t (n ) t (n) n n t (n ) (1 n n )
>
Solution 2: We can also solve the problem just using the components of the equation (A.6),
&
&
TS t (n) > : (n n )@n , i.e.:
TS i
>
t i(n) n i n k t (kn )
t i(n ) (n k n l Tkl ) n i
t (kn ) E ik n i n k t (kn)
t (kn ) E ik n i n k
(A.8)
where P is known as the Lagrange multiplier. Then, the derivative of the function L(n )
with respect to n and P , yields the following set of equations:
wL (n , P)
wn
wL (n , P)
wP
&
2 T sym n 2Pn 0
n n 1 0
&
( T sym P1) n 0
(A.9)
n n 1
The first set can only be solved if and only if det ( T sym P1) 0 , which is the eigenvalue
problem of the symmetric part of T . That is, the maximum and minimum of T N
correspond to the eigenvalues of T sym . Now, if we consider that T1sym , T2sym , T3sym , are the
eigenvalues of T sym , we can then restructure these such that:
TIsym ! TIIsym ! TIIIsym
(A.10)
Then, the maximum value of T N is defined by TIsym , and the minimum value by TIIIsym .
OBS.: When the nomenclature TI , TII , TIII are used to represent the eigenvalues,
TI ! TII ! TIII is implicit.
NOTE: As expected, the extreme values of T N relate to the symmetric part T sym because
the antisymmetric part plays no role in the normal component, i.e.
T N n T n n T sym n .
128
xc3
&
t (n)
T3
&
TN
&
TS
T2
T1
arbitrary plane
xc2
xc2
xc1
xc1
t i(n) n i
TN
Tij n j n i
T1n 12 T2 n 22 T3 n 32
>1,0,0@ TN
(A.11)
T1 .
&
t (n)
t i(n) t i(n) T N2
T N2
Tij Tik n j n k T N2
(A.12)
(A.13)
If we now ask what values of n i maximize the function TS2 , this problem is equivalent to
find extreme values of the function:
TS2 Pn i n i 1
F (n )
(A.14)
0j
wF (n )
wP
(A.15)
^
^
n ^T
2 T T n
`
`
T n P`
n 1 T12 2 T1 T1 n 12 T2 n 22 T3 n 32 P
n 2 T22 2 T2 T1 n 12 T2 n 22 T3 n 32 P
2
3
2
1
T2 n 22
2
3
(A.16)
with the constraint n i n i 1 . Then, the analytical solution of the previous set of equations
results in the following possible solutions:
n 1
solutions
n 2
n 3
TS
(1)
n 1(1)
r1
n (21)
n (31)
TS
(2)
n 1( 2 )
n (22 )
r1
n (32 )
TS
(3)
n 1(3)
n 1(3)
n 1(3)
r1
TS
(4)
(5)
(6)
2
1
1
2
1
2
0
129
TS
T T3
r 2
2
TS
T1 T3
2
TS
T1 T2
2
(A.17)
where the values of TS were obtained by substituting the values of n i into (A.13).
The first three sets of solutions give us the minimum values of TS , which correspond
precisely to the principal directions.
For solutions (4), (5) and (6) the planes are outlined as shown in Figure A.3, Figure A.4 and
Figure A.5, respectively.
Then, by ordering the eigenvalues (the principal values T1 , T2 , T3 ) such that
TI ! TII ! TIII we can find the absolute maximum tangential component:
TS max
TI TIII
2
(A.18)
>0
n i
1
2
1
2
T3
n i
>0
1
2
n i
>0
1
2
1
2
T2
T2
T1
1
2
T1
n i
>0
1
2
1
2
130
T3
>
n i
1
2
1
2
T3
>
n i
1
2
1
2
T2
T2
T1
>
n i
1
2
0
1
2
T1
>
n i
1
2
0
1
2
T3
>
n i
1
2
1
2
n i
n i
>
1
2
1
2
n i
>
1
2
1
2
1
2
1
2
@
T2
T2
T1
>
T1
131
The first example, (see Figure A.6), is a non-symmetric tensor and it is noteworthy that it is
&
positive definite since T N n T n ! 0 for all n z 0 . We will also verify that it has three
real eigenvalues, which correspond to TS 0 . Note, the maximum and minimum values
for the T N coincide with the eigenvalues of the symmetric part of T .
Then, for the tangential vector, we can carry out the following decomposition:
>
s
T
n @ s >s T
&
TS
sym
n s T skew n @s
(A.19)
TS
>s T
sym
n s T skew n @s
>s On s T
skew
n @s
>s T
skew
n @s
(A.20)
where we have considered that s n 0 , since the unit vectors s and n are orthogonal to
each other. So, the tangential component can be obtained as:
TS
& n )
s T skew n s t (skew
& n )
r t (skew
s s
& n )
r t (skew
(A.21)
We must emphasize that this procedure is only valid when n is a principal direction of
T sym , but not on an arbitrary plane.
For example, for the eigenvalue TIsym 10.55 , which is associated with the eigenvector
n (j1) > 0.45229371;0.561517458;0.692913086@ , we have:
& n )
r t (skew
TS
0 1 1 0.45229371
r 1 0 2 0.561517458
1 2 0 0.692913086
Tijanti
& n )
r t (skew
& n )
r t (skew
r2.424378
(A.22)
r0.55947
(A.23)
r2.41
(A.24)
n (j1)
3.61 , we find:
0 1 1 0.88542667
r 1 0 2 0.18949182
1 2 0 0.42439659
0.1313956
r 1.8381199
1.5753286
0.234905
r 0.036633496
0.50644304
0.84 , we obtain:
0 1 1 0.107004733
r 1 0 2 0.80547563
1 2 0 0.582888489
1.3883641
r 1.2727817
1.5039465
Note that the maximum and minimum normal components of T corresponds to the
eigenvalues of T sym , i.e., T N max T Isym 10.55 and T N min T Isym
0.84 . With respect to the
II
symmetric part, the maximum tangential component is equal to the radius of the circle
TIsym 10.55
and
TIsym
0.84 , (see Figure A.6), which is
formed by
II
TSsym max
10.55 0.84
2
4.86 .
132
TS
T N min
0.84; TS
TIII
T N min
2.41
1.52
0.84; TS
5 3 1
1 4 2
3 6 6
Tij
T N max
TII
TI
3.59
2.41
10.55; TS
9.89
T N max
10.55; TS
2.424
TN
2.424
TSsym
TSsym max
5 2 2
2 4 4
2 4 6
4.86
Tijsym
TIsym
II
0.84
TIsym
I
3.61
TIsym
10.55
TN
TI TIII
2
TI TIII
2
1.5 , TS
Intuitively, this leads us to believe that the graphical representation of a spherical tensor
reduces to a single point.
133
TS
TS max
2 .5
Tij
TII
TIII
4 0 0
0 1 0
0 0 1
TN
1
TI
T N max
Figure A.7: Graphical representation of a symmetric tensor with two equal eigenvalues.
The third example is a non-symmetric tensor, which has one real eigenvalue that is equal to
T1 0.964 , (see Figure A.8).
In the principal direction of the symmetric part of T , we have the following values for the
tangential component:
For the eigenvalue TIsym
TS
& n )
r t (skew
5 1.5 0.707427855
0
r 5 0 3 0.514420622
1.5 3 0 0.484682632
& n )
r t (skew
9.894 , we have:
3.299127
r 4.991187
0.482120
r6.0
(A.25)
2.02 , we have:
5 1.5 0.0575152387
0
r 5 0 3 0.72538138475
1.5 3 0 0.685940116897
2.5979967
r 2.3453965
2.2624170
r4.1676 (A.26)
134
TS
TS max
T N min
2.02; TS
TI
T N min
T N max
9.894; TS
6 .0
9.894; TS
6.0
4.17
TN
0.964
2.02; TS
Tij
6 8 4
2 2 1
1 7 2
4.17
T N max
TSsym
TSsym max
TIsym
II
2.02
Tijsym
TIsym
I
2.126
6 3 2.5
3 2 4
2.5 4 2
TN
TIsym
9.894
Figure A.8: Graphical representation of a non-symmetric tensor with only one real
eigenvalue.
& &
t (n) t (n)
TS2 T N2
&
t (n)
135
(A.27)
xc3
xc3
T III
TI
T II
TI
3
ec
T II
xc2
2
ec
1
ec
&
t (n)
B
xc2
T III
xc1
A
xc1
a)
b)
The components of the vector t (n) , in an arbitrary plane, were obtained in the Eq. (A.1).
&
TI n 1
;
t (2n) TII n 2
;
t 3(n) TIIIn 3
& &
Next, the dot product t (n) t (n) is evaluated as follows:
& &
t (n ) t ( n )
t (i n) e ci t (jn) e cj t i(n) t (jn) E ij t i(n) t i(n)
(A.28)
(A.29)
t 1(n)
2 2
TI2 n 12 TII2 n 22 TIII
n3
(A.30)
Then, the normal component T N , in the principal space, can be expressed as:
TN
&
t (n) n
Tij n j n i
TI n 12 TII n 22 TIIIn 32
(A.31)
TI (1 n 22 n 32 ) TII n 22 TIIIn 32
n 22
T N T III n 32 TI n 32 TI
( TII TI )
(A.32)
2 2
TI2 n 12 TII2 n 22 TIII
n3
2
2
2
2 2
TI (1 n 2 n 3 ) TII2 n 22 TIII
n3
(A.33)
136
and substituting n 22 , obtained in (A.32), into the above equation results in:
TS2 T N2
n 32
TI T N TII TIII
(A.34)
we obtain:
( T N TI )( T N TII ) TS2
( TIII TI )( TIII TII )
(A.35)
and we can find n 12 and n 22 in a similar fashion. Finally, we can summarize the results as
follows:
n 12
(a)
n 22
( T N TIII )( T N TI ) TS2
t0
( TII TIII )( TII TI )
(b)
n 32
( T N TI )( T N TII ) TS2
t0
( TIII TI )( TIII TII )
(c)
(A.36)
Then, if we consider that TI ! TII ! TIII we can verify that the equations (A.36) (a) and (c)
have positive denominators; so, their nominators must also be positive. However, the
equation in (A.36) (b) has negative denominators, so its nominator must be negative too.
i.e.:
n 12
n 22
n 32
>( T
2
( T N TIII )( T N TI ) T S d 0
t 0 ( T N TIII )( T N TI ) TS2 d 0
>( TII TIII )( TII TI )@ 0
( T N TI )( T N TII ) TS t 0
( T N TI )( T N TII ) T S2 t 0
t0
>( TIII TI )( TIII TII )@ ! 0
>
>
(A.37)
Then after some algebraic manipulations, the previous inequalities in (A.37) become:
TS2 >T N 12 ( TII TIII )@ t >12 ( TII TIII )@
2
TS2
(A.38)
The above shows equations of circles. The first circle with the center 12 ( TII TIII ) and
radius 12 ( TII TIII ) , indicates that the feasible points for the pair ( T N ; TS ) are outside the
, including the circumference, (see Figure A.10). The second circle with the
circle C1
1
center 2 ( TI TIII ) and radius 12 ( TI TIII ) , indicates that the feasible values for the pair
, including the circumference. The third inequality
( T N ; TS ) are inside the circle C 2
, which
indicates that the feasible values for the pair ( T N ; TS ) are outside the circle C 3
1
1
is defined by, the radius 2 ( TI TII ) and center 2 ( TI TII ) . Then, taking into account the
three equations simultaneously, the feasible region is formed by the pair ( T N ; TS ), defined
in the gray area which includes the circumferences C1 , C 2 and C 3 , (see Figure A.10).
137
feasible region
TS
TII
TIII
C3
TI
TN
C1
C2
TIII
Point N
Point Q
TII
TII
TII
TI
TIII
Point M
TI
Point Nc
TI
Point M c
Point Qc
TS
TS max
TS max
TI TIII
2
N
Q
TIII
TII
TI
Qc
Nc
Mc
T N max
TN
138
&
T n t (n )
t 1(n )
(n )
t 2
(n )
t
3
components
o
n 1
n 2
n 3
T1n 1
T2 n 2
T3n 3
t 1(n)
T1
t (2n)
T2
(A.39)
t 3(n)
T3
&
Our goal now is to define the surface, in the principal space, that describes the vector t (n)
for all possible values of n . Then, if we consider the constraint of n , i.e. n 12 n 22 n 32 1 ,
and substitute the n i equation given by (A.39) we obtain:
2
t 1(n)
T12
t (2n)
T22
t 3(n)
T32
(A.40)
which represents an ellipsoid in the principal space of T , (see Figure A.12). This surface
describes the feasible values for t 1 , t 2 and t 3 . When two eigenvalues are equal, the surface
becomes an ellipsoid of revolution, whereas when the three eigenvalues are equal, the
surface is a sphere, so, tensors that exhibit this property are called Spherical Tensors, and any
direction for these is a principal direction.
xc2 , t (2n)
xc1 , t 1(n)
T2
T1
n
&
t (n)
T3
xc3 , t (3n)
139
vector, TSoct .
&
T Noct - normal octahedral vector
&
TSoct - tangential octahedral vector
T3
& (n )
t
&
TSoct
T2
B
B
&
T Noct
OA
OB
T1
1
1
1
3
1
n i
B
B
OC
T2
T3
A
T1
Figure A.13: The octahedral plane and octahedral vector (principal space).
The octahedral vector can be expressed in terms of components in the form:
&
t (n )
&
T1
3
e 1c
T2
3
e c2
T3
3
e c3
(A.41)
&
&
T
e c e c
T
T
e c
t (n) n 1 e 1c 2 e c2 3 e c3 1 2 3
3
3 3
3
3
3
I
1
T1 T2 T3 1 Tii T Tm
3
3
3
(A.42)
140
&
where T Noct is called the octahedral normal component. Then, the magnitude of TSoct , called the
octahedral tangential component, is obtained as follows:
TSoct
>
& &
2
t (n) t (n) T Noct
1 2
1
2
T1 T22 T32 >T1 T2 T3 @
3
9
1
2 I T2 6 II T
9
(A.43)
1
3
T1 T2 2 T2 T3 2 T3 T1 2
1
3
(A.44)
2
II dev
3 T
T T T
dev 2
1
TSoct
dev 2
2
dev 2
3
(A.45)
T Noct
2
II dev
3 T
1
2 I T2 6 II T
3
TSoct
Tm
T T T
dev 2
1
dev 2
2
dev 2
3
Octahedral
(A.46)
normal component
Octahedral
tangential
component
(A.47)
Notice that the octahedral normal and tangential components are the same for the 8
octahedral planes, (see Figure A.14).
Then, let us consider the principal space once again, here represented by the orthonormal
basis ( e 1c , e c2 , e c3 ), (see Figure A.15). Now, we can plot the coordinates ( T1 , T 2 , T3 ) , which
are denoted by the P in Figure A.15.
Note, any plane normal to the straight line OA is an octahedral (deviatoric) plane and the
specific plane passing through the origin is denoted by 3 . Finally, the straight line OA is
called the spherical axis (or hydrostatic axis).
Let us now consider a deviatoric plane passing through the point P , which is denoted by
o
3c , (see Figure A.15). Then, we will define three vectors, namely, OP , OA and AP .
o
Next, the vector OP can be expressed in terms of the principal values of T as:
o
OP
(A.48)
T1e 1c T2 e c2 T3 e c3
o
OA
p OP n
3 Tm
3
T1e 1c T2 e c2 T3 e c3 e1c
3 Tm
T Noct
e c2
3
e c3
T1 T2 T3
3
(A.49)
3 T Noct
3 Tm
141
(A.50)
T3
&
T Noct
&
T Noct
&
T Noct
&
TSoct
T2
&
TSoct
&
TSoct
&
TSoct
&
T Noct
T1
Spherical axis
3c
T1
P ( T1 , T 2 , T 3 )
T1
T2
ATm , Tm , Tm
n i
3
1
ec
p
D
O
2
ec
n
D
T3
1
1
1
3
1
Deviatoric plane
3
ec
(Octahedral plane)
T3
T2
3 -plane
(deviatoric plane)
OA
OA n
e c e c
e c
3 Tm 1 2 3
3
3
3
Tm e 1c Tm e c2 Tm e c3
(A.51)
o
Then, the coordinates of point A are ( Tm , Tm , Tm ) and once the vectors OP and OA
o
have been defined, the vector AP can be evaluated by adding the following vectors, (see
Figure A.15):
142
(A.52)
AP OP OA
Then, taking into account (A.48) and (A.51), the above equation becomes:
T1 e 1c T2 e c2 T3 e c3 Tm e 1c Tm e c2 Tm e c3
( T1 Tm )e 1c ( T2 Tm )e c2 ( T3 Tm )e c3
AP
T1dev e 1c
T2dev e c2
(A.53)
T3dev e c3
Tijdev
o
(A.54)
AP
Notice that, the components of AP represent the principal values of the deviatoric part
o
T T T
dev 2
1
AP
dev 2
2
dev 2
3
(A.55)
( T1dev ) 2 ( T2dev ) 2 ( T3dev ) 2 ,
3 TSoct
(A.56)
Note, we could also have obtained the AP module by using Pythagoras theorem:
o
AP
OP OA
1
T1 T2 T3 2
3
2 2
T1 T22 T32 T1 T2 T2 T3 T3 T1
3
AP
(A.57)
(A.58)
where q indicates how faraway is the tensor state is from the spherical state, (see Figure
A.15).
Note, in some cases working in the 3 -plane could be advantageous and for this reason we
define some parameters on it.
We will next analyze the projection of the principal space onto the 3 -plane, (see Figure
A.16).
Then, to find the components of the unit vector e 1cc a1e 1c a 2 e c2 a 3 e c3 described in
Figure A.16, we will consider the principal space as shown in Figure A.17, where
cos C
sin B
2
3
a1 , a 2
a1e 1c a 2 e c2 a3 e c3
a 2 a3
a1
2
3
1
e1c e c2 e c3 0 1 a1 a 2 a3 0
3
3
2
6
(A.59)
a2
1
a1
2
a3
3c , 3
2
2 3
(A.60)
Tc1
T1
Qc
P
T
143
cos B
1
ec
e 1cc q cos T
1
3
e c2 , e c3
120
T2 , T3
Tc2
Tc3
T1
n
2
1
ec
e 1cc
C B
3
ec
3
T2
2
ec
1
6
( 2e 1c e c2 e c3 )
(A.61)
1
6
( 2e 1c e c2 e c3 )
(A.62)
0 T1dev
144
q cos T OP e 1cc
(2 T1dev T1dev )
(A.63)
2 II T dev , we obtain:
3 dev
T1 2 II T dev cos T
2
q cos T
3 dev
T1
2
T1dev
3
2
cos T
T1dev
3 dev
T1
2
II T dev
T1dev
(A.64)
II T dev cos T
Likewise, we can find T2dev and T3dev . Then, we can express the principal values,
Tij
T1
0
Tm E ij Tijdev , as follows:
0
T2
0
0
0
T3
Tm
0
0 T1dev
0 0
Tm 0
0
Tm
0
Tm
0
0
T3dev
0
T2dev
0
0
2
0
II T dev
3
Tm
0
Tm
0
cos T
0
cos(T
0
0
0
2S
)
0
3
2S
cos(T 3 )
0
(A.65)
0
T2
0
0
0
T3
cos T
0
0
p 0 0
1
2 q 0
2S
0
0
cos(
)
0
p
T
3
3
3
0
0 0 p
0
cos(T 23S )
Spherical
part
(A.66)
Deviatoric
part
Then, substituting the cos T , given by the equation in (A.64), into the trigonometric
relationship cos 3T 4 cos 3 T 3 cos T , yields:
3
cos 3T 4
2
T1dev
II T dev
3 3
T1dev
II T dev
cos 3T
3 3
2 II T dev
3 3
2 II T dev
T dev
1
T1dev II T dev
dev
3 3 T1
2 II
T
dev 3
2
T1dev T2dev T3dev T1dev T2dev T3dev
T1
J3 III T dev
T1dev
cos 3T
(A.67)
3 3 III T dev
2 II T dev
Note, the terms II T dev , III T dev are invariants, hence cos 3T is an invariant too.
(A.68)
(A.69)
2 Continuum Kinematics
2
Continuum Kinematics
2.1 Introduction
A material body (continuum) in motion, starting from an initial state (t 0) { t 0 , will have
different configurations over time, (see Figure 2.1). In this Chapter we will study the
description of motion also called Kinematics, thus establishing the equations of motion that allow
us to characterize how the continuum evolves and how continuum properties, e.g.
displacement, velocity, acceleration, mass density, temperature, etc., change over time. To
do this, we will consider the initial configuration, also known as the reference or undeformed
configuration, characterized by material body B0 at time t 0 , and we will also consider the
generic configuration Bt at time t called the current configuration, also known as the actual or
deformed configuration.
Configuration at t1
Reference
Configuration - t 0
Current Configuration - t
Bt
B1
B0
K
145
146
S0 (X )
"im
'V0 o0
'm
'V 0
dm
dV0
kg
m3
(2.1)
Starting from this concept we can define a particle as a dimensionless element that has
physical properties such as mass density, velocity, temperature, etc.
0) { t 0
(t
Reference
configuration
'V0 o 0
&
S0 ( X )
B0
X3
X2
X1
S, v , T
at t 2
t1
t0
Pc
t3
- Current position
P ccc
t2
P cc
Trajectory of particle P
2 CONTINUUM KINEMATICS
2.2.1
147
Kinds of Motion
As we have seen previously, in rigid body motion the distances between particles remain
unchanged. Then we can establish an equation that governs this motion. To do this let us
consider a Cartesian system OX 1 X 2 X 3 which is attached to the body, so, the position
vector of any particle with respect to this system remains unchanged during motion. We
can also adopt a second Cartesian system ox1 x 2 x3 , which is represented by the
orthonormal basis ( e 1 , e 2 , e 3 ), (see Figure 2.4).
&
x
&
&
c (t ) Q(t ) X
time - t
x3
X3
t0
&
x
e 3
o
e 1
&
c(t )
e 2
&
X
X2
I 2
I 3
O I
1
X1
x2
x1
&
If X and x are the position vectors of material point P with respect to the systems e i
& &
cX
(2.2)
&
where c(t ) is a time-dependent vector that describes the translation motion of the system
I . The above equation (2.2) in symbolic notation can be represented as:
i
x p e p
c k e k X j I j
(2.3)
148
The components of (2.3) in the system ox1 x 2 x3 are obtained by means of the dot product
with respect to the basis e i , i.e.:
x p e p e i
x p E pi
xi
in where I j e i
c k e k e i X j I j e i
c k E ik X j a ji
(2.4)
c i X j a ji
and where it holds that a ik a kj E ij , i.e. a ji is an orthogonal matrix. Note also that the
equation (2.4) holds for any adopted system. If we consider that Q ij a ji , where Q ij are
orthogonal tensor components, we can sum up the equation in (2.4) as:
&
x
&
&
c Q X
(2.5)
Cc
Dc
30
b
A
xc1
D
Bc
30
Ac
X 1 , x1
&
&
&
&
&
&
cos T sin T 0
sin T cos T 0
0
0
1
2
0
0
1 X 3
A particle which initially was at point D ( X 1 0 , X 2 b , X 3 0 ) moves into the following
x1
x2
x
3
position:
2 CONTINUUM KINEMATICS
x1D
D
x2
x D
3
149
0
0
1 0
0
&
In the above example we have adopted the system, X , fixed in space and time. When we
are establishing the equations of motion for a continuum subjected to deformation, we can
adopt a system fixed in space and time which is called the material system. There is also a
&
second system defined by x and is called the spatial system.
&
&
In general, throughout this chapter we will adopt c 0 , (see Figure 2.5), or to put it
another way, the spatial and the material axes will be superimposed as shown in Figure 2.5.
X 3 , x3
I 3 , e 3
O
I 1 , e 1
I 2 , e 2
X 2 , x2
X 1 , x1
2.2.2
Types of Configurations
We define two types of configurations adopted in this chapter, (see Figure 2.6), namely:
The Reference configuration or the initial configuration: the configuration at the instant of
time (t 0) { t 0 , is considered to be the undeformed configuration in which the
&
particle P is identified by the position vector X P .
The Current configuration or deformed configuration: the configuration at the instant of time
t.
150
K
x3 , X 3
Reference
configuration - t 0
B0
I 1 , e 1
&
Bt
&
S( x, t )
Pc
S0 (X )
I 3 , e 3
O
Trajectory of particle - P
K 1
I 2 , e 2
x2 , X 2
Current configuration - t
x1 , X 1
Mass Density
As with the definition of mass density in the reference configuration given in (2.1), we
define mass density in the current configuration, (see Figure 2.7), as:
'm
'V
dm
dV
kg
(2.6)
m3
&
Mass density is a scalar field, which is a function of position and time, i.e. S S ( x, t ) .
"im
'V o 0
K
(t
Initial configuration
0) { t 0
'V0 o 0
&
S0 ( X )
'
V
o
0
&
S ( x ,t )
x3
Current configuration
x1
x2
2 CONTINUUM KINEMATICS
151
Let us consider the material body B0 in the initial configuration, (see Figure 2.8). At an
arbitrary time (t ) , the material body occupies a new position in space, Bt . We now focus
our attention on a particle of the continuum, which is denoted by the particle P .
NOTE: With regard to nomenclature, the particles are identified by labels. These labels are
the positions they occupied in the reference configuration. For instance, the particle which
occupied the point P ( X 1 , X 2 , X 3 ) in the reference configuration will be denoted by
particle P , (see Figure 2.8).
Trajectory of particle P
Initial configuration - t 0
x3 , X 3
Bt
&
u
B0 P
Pc
&
x
&
X
Current configuration - t
I 3 , e 3
O
I 1 , e 1
I 2 , e 2
x2 , X 2
x1 , X 1
X 1e 1 X 2 e 2 X 3 e 3
(2.7)
X1
X
2
X 3
(2.8)
x1e 1 x 2 e 2 x3 e 3
(2.9)
152
xi
2.3.2
x1
x
2
x3
(2.10)
2.3.3
& &
xX
ui
>m@
xi X i
(2.11)
The velocity of a particle is defined by the rate of change of the position vector, i.e.:
&
V
2.3.4
&
dx &
{x
dt
& &
d u X
dt
&
&
du dX
dt ,
dt
&
0
&
du &
{u
dt
m
s
(2.12)
m
s2
(2.13)
2.3.5
&
dV &
{V
dt
&
d 2 x & &
{x u
dt 2
Continuum properties, e.g.: mass density, temperature, velocity, acceleration, etc., are
intrinsic in particles (material points), and such properties may change over time. As
mentioned before, continuum motion is characterized by the bijective function ( K ), and
the inverse function ( K 1 ). This ensures that we can correlate continuum properties
between the current and reference configurations. In other words, the study of motion can
be carried out either in the current or reference configuration.
2.3.5.1
The particle in motion can be described in terms of material coordinates ( X ) and time as:
&
x
& &
x X,t
xi
&
x i X 1, X 2 , X 3 , t xi X , t
(2.14)
The equations of motion (2.14) are called the Lagrangian or Material Description of the motion.
&
The above parametric equation gives us the current position x , at time t , of a particle that
&
occupied position X in the reference configuration, at time t 0 . The equation in (2.14),
applied to particle P , provides us with the unique path line (trajectory) of the particle, (see
Figure 2.9).
2 CONTINUUM KINEMATICS
2.3.5.2
153
Particle motion can also be described in terms of spatial coordinates ( x ) and time as:
&
X
& &
X x, t
Xi
&
X i x 1 , x 2 , x3 , t X i x , t
(2.15)
&
The above equation give us the original position X , at time t 0 , of a particle which at the
present time ( t ) has the coordinates ( x 1 , x 2 , x3 ) , (see Figure 2.9).
wxi
wX j
wx1
wX 1
wx 2
wX 1
wx3
wX 1
wx1
wX 2
wx 2
wX 2
wx3
wX 2
wx1
wX 3
wx 2
z0
wX 3
wx3
wX 3
m3
3
m
(2.16)
where J is called the Jacobian determinant, also knows as the volume ration.
OBS.: The Axiom of Impenetrability: Two particles can not occupy the same place at
the same time. As discussed later, this condition is ensured when the Jacobian
determinant is positive, i.e. J ! 0 .
& &
x P X,t
and
& &
&
x X P , t0 x P
&
&
&
x X P , t1 x P c
& &P
&
x X , t 2 x Pcc
Trajectory of particle P
2.3.5.3
&
XP
(2.17)
& &
&
X x P , t0 X P
& &
&
X x S c , t1 X S
& &
&
X x Qcc , t
XQ
2
(2.18)
Particles at point P,
at different time
(2.19)
NOTE: Some authors try to differentiate a Lagrangian from Eulerian variable by using
upper and lower-case letters, respectively. As a general rule we have not adopted this
convention in this publication. In this textbook when we are dealing with a Lagrangian
& & &
variable we will indicate explicitly by its arguments, i.e. V { v ( X , t ) . And if we are dealing
& &
with an Eulerian variable it will be indicated as follows v ( x , t ) .
154
& &
X (x P , t2 )
& &
x( X P , t2 )
&
XQ
&
x Pcc
t2
point P { Q cc
Particle Q
Trajectory of particle P
S cc
&
x Qcc
t1
& &
X ( x S c , t1 )
&
x Pcc
&
XS
Pcc
Particle P
Qc
Sc { P
&
x Sc
t0
Pc
&
x Pc
&
& &
X ( x P { S c , t1 ) X S
&
& &
x ( X P , t1 ) x P c
Particle P
&
S P
XQ
&S
X &
&
X P xP
Reference configuration
at t
&
X
&
t0 x
Matrix form
o x 2
x 2 ( X , t ) X 3 t X 2
&
x
3
x 3 ( X , t ) X 3
1 t 2
0 1
0 0
0 X 1
t X 2
1 X 3
(2.20)
Is the motion above possible? If so, find the displacement, velocity and acceleration fields
in Lagrangian and Eulerian descriptions. Consider a particle P that at time t 0 was at the
point defined by the triple equation X 1 2, X 2 1, X 3 3 . Find the velocity of P at time
t 1s and t 2 s .
Solution: Motion is possible if J z 0 , thus
wxi
wX j
wx1
wX 1
wx 2
wX 1
wx3
wX 1
wx1
wX 2
wx 2
wX 2
wx3
wX 2
wx1
wX 3
wx 2
wX 3
wx3
wX 3
1 t2
0
0
t
1
1
0
1z 0
&
&
&
The displacement vector field is given by the definition in (2.11), u x X . Using the
equations of motion (2.20) we obtain:
u1
u 2
u 3
X&& , t
X& , t
X , t
&
x1 ( X , t ) X 1
&
x2 ( X , t) X 2
&
x3 ( X , t ) X 3
>X
2t
X1 X1
>X 3 t X 2 @ X 2
>X 3 @ X 3 0
X 2t 2
X 3t
(2.21)
2 CONTINUUM KINEMATICS
155
which are the components of the displacement vector in the Lagrangian description. Here,
velocity and acceleration can be evaluated as follows:
&
&
d u1 X , t
V1 { v1 ( X , t )
dt
&
&
du 2 X , t
{
(
,
)
V
v
t
X
2
2
dt
&
&
du 3 X , t
V3 { v 3 ( X , t )
dt
d
X 2t 2
dt
d
X 3t
dt
2 X 2t
X3
d
X 2t 0
dt
&
dV1
2X 2
A1 { a1 ( X , t ) dt
&
dV 2
0
A2 { a 2 ( X , t )
dt
&
dV 3
0
A3 { a 3 ( X , t ) dt
(2.22)
The inverse form of (2.20) provides us the equations of motion in the Eulerian description:
1 t 2
1
0
0
0
X1
X 2
X
3
&
t 3 x1 X 1 ( x , t ) x1 t 2 x 2 t 3 x 3
&
t x 2 X 2 ( x , t ) x 2 tx 3
&
1 x 3 X 3 ( x , t ) x 3
(2.23)
Then, the displacement, velocity and acceleration fields in Eulerian description can be
evaluated by substituting equation (2.23) into the equations (2.21) and (2.22), i.e.:
X& ( x& , t ), t
X& ( x& , t ), t
X ( x& , t ), t
& &
V X ( x , t ), t
& &
V X ( x , t ), t
& &
V X ( x , t ), t
& &
A X ( x , t ), t
& &
A X ( x , t ), t
& &
A X ( x , t ), t
u1
u 2
u 3
&
&
&
X 2 ( x , t )t 2 ( x 2 tx 3 )t 2 u1 ( x , t )
&
&
X 3 ( x , t )t x 3 t u 2 ( x , t )
&
u3 ( x, t ) 0
&
&
2 X 2 ( x , t )t 2( x 2 tx 3 )t v1 ( x , t )
&
&
X 3 ( x, t ) x3 v2 ( x, t )
&
v3 ( x, t ) 0
&
&
2 X 2 ( x , t ) 2( x 2 tx 3 ) a1 ( x , t )
&
a2 ( x, t ) 0
&
a3 ( x, t ) 0
(2.24)
(2.25)
(2.26)
Taking into account the Lagrangian description of velocity given in (2.22), the velocity of
particle P ( X 1 2, X 2 1, X 3 3 ) at time t 1s is given by:
&
v1 ( X , t )
&
&
2 m / s ; v 2 ( X , t ) X 3 3m / s ; v 3 ( X , t ) 0
We can also observe that at time t 1s the particle P occupies the position:
x1 X 2 t 2 X 1 3 ; x 2 X 3 t X 2 4 ; x 3 X 3 3
So, the velocity of the particle P can also be evaluated by (2.25) as:
&
v1 ( x , t ) 2( x 2 tx 3 )t 2( 4 1 u 3) u 1 2m / s
&
v 2 ( x , t ) x 3 3m / s
&
v ( x , t ) 0
3
2 X 2t
Note that, the velocities obtained via the Lagrangian or Eulerian description are the same,
since velocity is an intrinsic property of the particle.
We can also provide the velocity of the particle P at time t 2 s :
At time t
&
V1 { v1 X , t 2 X 2 t 2 u 2 u 1 4 m / s
&
V 2 { v 2 X , t X 3 3m / s
&
V3 { v 3 X , t 0
2 s the new position of P is:
156
&
x1 ( X , t ) X 2 t 2 X 1 6
&
x2 ( X , t ) X 3t X 2 7
&
x 3 ( X , t ) X 3 3
& &
As we can verify the Lagrangian description of motion x ( X , t ) describes the trajectory of
P.
Trajectory of particle P
&
Vi P ( X P , t 1s)
t0
X iP
>2;1;3@
&
viP ( x , t 1s)
>2;3;0@
>2;3;0@
&
Vi ( X P , t
P
t 1s
2s )
xiP
>4;3;0@
>3;4;3@
P
xiP
>6;7;3@
2s
&
viP ( x , t
2s )
>4;3;0@
(2.27)
&
dT( X , t )
dt
(2.28)
When the property is described in terms of material coordinates, this implication is that the
property represented is connected to the same particle during motion.
2 CONTINUUM KINEMATICS
t0
particle trajectory
T0
t1
t2
particle trajectory
particle trajectory
P
&
X
T1
P
& &
x ( X ,0)
157
( X1, X 2 , X 3 )
Pc
& &
x ( X , t1 )
T2
( X1, X 2 , X 3 )
& &
x ( X , t2 )
Pcc
(2.29)
In this description the observer is not traveling with the particle, but fixed at one point
( x1 , x 2 , x 3 ) watching the particles passing. According to Figure 2.11, equation (2.29), at
time t1 , provides us with the property of particle Q , which takes the point ( x1 , x 2 , x 3 ) .
Later, t 2 , in equation (2.29) give us the property of another particle, e.g. R , and at time t 3 ,
the equation (2.29) give us the value of the property of particle P , (see Figure 2.11).
t1
t2
Trajectory of particle Q
trajectory of particle R
trajectory of particle P
R
R
&
( x)
t3
&
wT ( x , t1 )
wt
Q
&
( x)
Q
&
wT R ( x , t2 )
wt
&
( x)
Q
&
wT P ( x , t3 )
wt
So,
158
&
&
&
wT( x , t ) wT( x, t ) wx k ( X , t )
wt
wt
xk
&
&
DT( x , t )
T ( x , t ) {
Dt
local
rate of change
(2.30)
Convective
rate of change
&
&
&
wT( x, t ) wT( x , t )
vk ( X , t )
wt
wx k
& &
& &
& &
Then, we can define the material time derivative operator for an Eulerian property, x ( x , t ) ,
as:
&
D( x , t )
Dt
2.4.1
&
&
& &
w( x , t )
Material time derivative
x& ( x , t ) v ( x , t )
for an Eulerian variable
wt
&
D( x , t )
Dt
&
&
w( x , t ) w( x , t )
vk
wx k
wt
(2.31)
(2.32)
D & &
x( X , t )
Dt
&
x P
&
dx
dt x
xP
& &
d u X
dt
& &
du( X , t )
dt
(2.33)
& &
&
D2 & &
x P { 2 x ( X , t )
Dt
(2.34)
The Eulerian acceleration can be evaluated either by substituting the inverse equation of
motion into the equation in (2.34) or via the definition of the material time derivative for
an Eulerian property, i.e.:
&
a iP ( x , t )
&
D
vi ( x, t )
Dt
&
&
wvi ( x , t ) wv i ( x , t ) wx k
wt
wx k
wt
&
&
wvi ( x , t ) wvi ( x , t )
&
v k ( x, t )
wt
wx k
convective
acceleration
& &
& &
Dv ( x , t )
a ( x, t ) {
Dt
& &
& & &
wv ( x , t )
x& v v ( x , t )
wt
(2.35)
2 CONTINUUM KINEMATICS
a1
&
a i ( x , t ) a 2
a
3
wv1 wv1
wt wx1
wv wv
2 2
wt wx1
w
v
3 wv3
wt wx
1
wv1
wx 2
wv 2
wx 2
wv3
wx 2
159
wv1
wx3 v
1
wv 2
v
2
wx3
v 3
wv3
wx3
(2.36)
Returning to Problem 2.2, the Eulerian velocity field was obtained as:
&
v1 ( x , t )
&
2( x 2 tx 3 )t ; v 2 ( x , t )
x3
&
; v3 ( x , t )
(2.37)
Explicitly, the Eulerian acceleration can also be evaluated by using the definition in (2.35):
&
a iP ( x , t )
&
&
wvi ( x, t ) wvi ( x , t )
&
v k ( x, t )
wt
wx k
&
&
&
&
wvi ( x, t ) wv i ( x , t )
wv ( x , t )
wv ( x, t )
&
&
&
v1 ( x , t ) i
v 2 ( x, t ) i
v 3 ( x , t )
wt
wx 2
wx3
wx1
(2.38)
&
&
&
&
wv ( x , t )
wv ( x , t )
wv1 ( x , t ) wv1 ( x , t )
&
&
&
v3 ( x , t )
v 2 ( x, t ) 1
v1 ( x , t ) 1
w
w
w
wt
x
x
x
1
2
3
2( x 2 x 3 t )
&
&
&
&
wv ( x , t )
wv 2 ( x , t ) wv 2 ( x , t )
wv ( x , t )
&
&
&
&
a 2P ( x , t )
v1 ( x , t ) 2
v 2 ( x, t ) 2
v 3 ( x , t )
w
wt
w
w
x
x
x
1
2
3
(2.39)
0
&
a 3P ( x , t )
&
&
&
&
wv ( x , t )
wv ( x , t )
wv3 ( x , t ) wv3 ( x , t )
&
&
&
v 2 ( x, t ) 3
v3 ( x , t )
v1 ( x , t ) 3
wx 2
wx3
wt
wx1
2.4.2
Stationary Fields
&
A field G ( x , t ) is said to be stationary if the local rate of change does not vary over time:
&
wG ( x , t )
wt
&
(2.40)
For example, let us consider a stationary (steady state) velocity field as shown in Figure
2.12. Then, as we can verify, the field representation for any time, e.g. t1 and t 2 , does not
change. However, that does not mean that the velocities of the particles do not change
over time. In light of Figure 2.12, we can now focus our attention on the fixed spatial point
&
&
&
x * . At time t1 the particle Q is passing through point x * with velocity v * . Let us also
consider another particle P , which is passing through another point with velocity
&
&
&
v P (t1 ) z v * . At time t 2 the particle P is now passing through the point x * . It follows that
&
if we are dealing with a steady state velocity field, then the velocity of particle P at x *
&*
&P
&*
must be v , i.e. v (t 2 ) v . We can easily contrast this with the material time derivative of
velocity, which is always associated with the same particle, i.e.:
160
& &
Dv ( x , t ) & &
{ a ( x, t )
Dt
& &
& & &
& & &
wv ( x , t )
x& v v ( x ) x& v v ( x )
wt
& &
a ( x)
(2.41)
&
0 (Stationay )
The rate of change of velocity (acceleration) will be zero if the velocity field is stationary
& &
wv ( x , t)
wt
&
&
0 and homogeneous ( x& v
0 ).
We can also verify that, although spatial velocity is independent of time, that does not
mean material velocity is also, since:
& & & & &
& &
v ( x ) v ( x ( X , t )) v ( X , t )
t1
(2.42)
& &
v ( x)
&
& &
v ( x * , t1 ) v *
Particle - Q
Particle- P
&
&
v P z v*
&
vQ
&
x*
t2
& &
v ( x)
&
& &
v ( x * , t2 ) v *
&
vP
Particle - P
&
x*
2 CONTINUUM KINEMATICS
2.4.3
161
Streamlines
Given a spatial velocity field at time t , we can define a streamline to the curve in which the
tangent at each point has the same direction as the velocity. In general, the streamline and
the trajectory do not coincide, but in steady state motion they do.
& &
a ( x , t)
&
v2
wv
x&
wt
2
&
&
&
v ( x& v )
&
& &
wv
x& v v
wt
&
v2
wv
x&
wt
2
&
&
v rot v
&
v2
wv
x&
wt
2
& &
rot v v
Solution:
To prove the above relationship one need only demonstrate that:
&
&
&
v ( x& v )
v2
x&
2
& &
x& v v
Expressing the terms on the right of the equation in symbolic notation we obtain:
v2
x&
2
&
&
&
v ( x& v )
1 w
v j v j v i e i
e r v s e s
e i
2 wx i
wx r
Using the definition of the permutation symbol (see Chapter 1) we can express the vector
product as:
v2
x&
2
&
&
&
v ( x& v )
v2
x&
2
&
&
&
v ( x& v )
wv
1 w
v j v j v i e i . rst s e t
e i
2 wx i
wx r
wv j
wv s
1
e k
e i 2v j
. rst . itk v i
2
wx r
wx i
where we have used the equation e i e t . itk e k . In Chapter 1 we also proved that
. rst . itk . rst . kit E rk E si E ri E sk , then:
&
wv j
v2 &
wv
&
vj
e i E rk E si E ri E sk v i s e k
x& v ( x& v )
wx i
wx r
2
wv j
wv
wv
vj
e i E rk E si v i s E ri E sk v i s e k
x
wx i
w
wx r
r
wv j
wv s
wv k
e k
e i v s
vj
vi
wx i
wx i
wx k
wv j
wv
wv
e i v s s e k v i k e k
wx i
wx k
wx i
wv j
wv s
wv
e i v s
E sj v s
E ik e i v i k e k
wx i
wx k
wx i
wv s
wv s
wv k
e i v s
e i v i
e k
vs
wx i
wx i
wx i
&
wv k e k
w (v )
vi
vi
wx i
wx i
&
&
x& v v
vj
162
& &
&
Problem 2.4: Consider the equations of motion x ( X , t ) and the temperature field T ( x , t )
given by:
x1
x2
x
3
X 1 (1 t )
X 2 (1 t )
&
T ( x)
x12 x 22
X3
Find the rate of change of temperature for the particle P at time t 1s given that particle
P was at point ( X 1 3, X 2 1, X 3 0) at time t 0 .
Solution 1:
In this first solution we first obtain the material time derivative of the Lagrangian
&
temperature, so, we have to obtain the temperature in Lagrangian description T ( X , t )
(Lagrangian temperature):
&
T ( x)
&
T ( X , t)
x12 x 22
p
By substi tuting
the equation of motion
p
X 12 (1 t ) 2 X 22 (1 t ) 2
&
&
DT dT ( X , t )
2 X 12 (1 t ) 2 X 22 (1 t )
T ( X , t ) {
dt
Dt
By substituting t 1s , ( X 1 3, X 2 1, X 3 0) , into the above equation we obtain:
&
T ( X , t ) 2 X 12 (1 t ) 2 X 22 (1 t ) 2(3) 2 (1 1) 2(1) 2 (1 1) 40
Solution 2:
In this second solution we directly use the definition of material time derivative of the
&
Eulerian variable, (see Eq. (2.30)). Then T ( x , t )
DT
Dt
&
&
&
wT ( x ) wT ( x )
vk ( x, t ) .
wt
wx k
x2
x
3
&
v1 ( X , t ) X 1
&
o v 2 ( X , t ) X 2
X 2 (1 t ) velocity
&
X3
v 3 ( X , t ) 0
X 1 (1 t )
x2
x
3
X1
X 1 (1 t )
X3
X 3
x1
(1 t )
x2
(1 t )
x3
V1 ( X ( x , t ), t ) X 1 ( x , t ) (1 t ) v1 ( x , t )
& &
x2
&
&
v 2 ( x, t )
V 2 ( X ( x , t ), t ) X 2 ( x , t )
(
1
)
t
V3 v 3 ( x& , t ) 0
2 CONTINUUM KINEMATICS
163
&
Afterwards, the material time derivative of the Eulerian temperature, T ( x , t ) , is given by:
&
DT ( x , t ) &
{ T ( x, t )
Dt
&
wT ( x )
wt
wT
wT
wT
v1
v2
v3
x
x
x3
w
w
w
2
1
0 (Stationar y field)
x
x
2 x1 1 2 x 2 2
1 t
1 t
The position of particle P at time t
x1
x2
x
3
&
T ( x , t )
0
&
T ( x , t )
2 x12 2 x 22
1 t 1 t
1s is evaluated as follows:
X 1 (1 t ) 3(1 1) 6
X 2 (1 t ) 1(1 1)
X3
2
( x12 x 22 )
1 t
Then, by substituting the spatial coordinates in the expression of the material time
derivative of temperature we obtain:
2
2
2, x 3 0, t 1)
( x12 x 22 )
(6 2 2 2 )
1 t
11
&
Alternatively, the expression T ( x , t ) could also have been obtained as:
&
T ( X , t ) 2 X 2 (1 t ) 2 X 2 (1 t )
&
T ( x , t ) T ( x1
6, x 2
& &
T ( X ( x , t ), t )
40
& 2
& 2
2> X 1 ( x , t )@ (1 t ) 2>X 2 ( x , t )@ (1 t )
x
x
2 1 (1 t ) 2 2 (1 t )
(1 t )
(1 t )
&
2
( x12 x 22 ) T ( x , t )
(1 t )
Introduction
In the previous section we studied the description of a particle in motion without looking
at how the relative motion between particles changed. In this section we analyze how
distances between particles change during motion after which we define some deformation
and strain tensors. To do this, let us consider two neighboring particles in the reference
configuration, which are denoted by P and Q , (see Figure 2.13).
2.5.2
&
Let dX be a vector joining two points P and Q in the reference configuration, defining a
&
line element. The unit vector associated with the dX -direction is represented by M . After
motion, particles P and Q occupy new positions Pc and Qc , respectively. In this new
configuration (current configuration), the vector joining the points Pc and Qc is
&
represented by dx , which is associated with the unit vector m , (see Figure 2.13), and the
&
&
magnitudes of dX and dx are denoted, respectively, by:
o
PQ
&
dX
dS
P cQ c
&
dx
ds
(2.43)
164
&
dx
Reference
configuration - t
M
Q
X 3 , x3
B0
&
XQ
&
&
F ( X , t ) dX
&
dX
&
dx
I 1 , e 1
&
xQ
Pc
Current
configuration - t
& &
x { xP
I 2 , e 2
ds
Bt
&
dx
I 3 , e 3
dS
&
dx
PcQc
Qc
&
&
XP { X
&
dX
&
dX
PQ
t1
X 2 , x2
X 1 , x1
&
dx
&
dX
ds
dS
O m ! 0 Stretch
where
(2.44)
ds o0
1 - There is no elongation;
&
&
dx dX
&
dX
ds dS
dS
Unit extension
(2.45)
The possible values of the unit extension are within the range of 1 F m f . The stretch
is related to the unit extension by:
F m
ds dS
dS
ds
1 O m 1
dS
ds (F m 1)dS
O m dS
(2.46)
2 CONTINUUM KINEMATICS
2.5.3
165
&
Our goal now is to find the relationship between the line elements dX and dx . By
& & &
considering the material description of motion x x ( X , t ) , and by applying vector
addition, (see Figure 2.13), we obtain:
&
XQ
&
&
& & &
& &
(2.47)
X P dX
;
dx x Q ( X Q , t ) x P ( X P , t )
&
&
&
&
&
&
&
&
&
If we observe that x Q ( X Q , t ) x P ( X P dX , t ) x ( X dX , t ) , the vector field dx in the
&
& &
& &
x ( X dX , t ) x ( X , t )
>xi ( X 1 dX 1 , X 2 dX 2 , X 3 dX 3 , t ) xi ( X 1 , X 2 , X 3 , t )@ e i
(2.48)
& 2
wxi
wx
wx
&
dX 1 i dX 2 i dX 3 e i O( dX ) dx
wX 3
wX 2
wX 1
& 2
wxi
dX j e i O( dX ) (2.49)
wX j
Since points P and Q are close together, higher order terms can be discarded, which leads
to:
&
dx
wxi
dX k e i
wX k
Fik dX k e i
(2.50)
&
dx
(2.51)
where F is a two-point tensor and is known as the material deformation gradient or simply the
&
deformation gradient. The relation in (2.51) is a linear transformation so F relates dX
&
(undeformed configuration) to dx (deformed configuration), (see Figure 2.13). The
equation in (2.51) could have been obtained by directly starting from the gradient
&
definition, (see Chapter 1-Tensors). That is, if G G( x , t ) is a scalar field, the total
&
&
& wG( x , t ) &
derivative ( dG ) is given by the equation: dG ( x , t ) G dx
& dx . Then, if we have
wx
& &
x ( X , t ) , the total derivative (differential) is:
& &
&
&
& &
& wx ( X , t ) &
(2.52)
dx
& dX X& x ( X , t ) dX F dX
wX
&
The components of dx in Cartesian system can be evaluated by means of the scalar
&
&
( dx ) j
&
dx e j
Fik dX k e i e j
F jk dX k
E ij
(2.53)
&
&
&
wx
wx
wx
e1
e2
e3
wX 2
wX 3
wX 1
wx i
e i e j
wX j
x i , J e i e j
(2.54)
166
Fij
wxi
wX j
xi , J
wx1
wX 1
wx 2
wX
1
wx3
wX 1
wx1
wX 3
wx 2
wX 3
wx3
wX 3
wx1
wX 2
wx 2
wX 2
wx 3
wX 2
(2.55)
In
this
publication,
w xi
w xi
z xi, j {
.
wX j
wx j
we
denote
the
material
w (x)
e i and the spatial gradient by grad(x) { x& (x)
wX i
gradient
w (x)
ei .
wxi
by
We can also find the inverse transformation of the equation in (2.51) i.e.:
&
dX
&
F 1 dx
(2.56)
F 1
Fij1
wX i
wx j
X I, j
(2.57)
Fij1
wX i
wx j
F 1
wX 1
wx1
wX 2
wx
1
wX 3
wx1
wX 1
wx 2
wX 2
wx 2
wX 3
wx 2
wX 1
wx 3
wX 2
wx 3
wX 3
wx 3
(2.58)
We can now show, (see Chapter 1), that the following relationships are valid:
F >cof (JF )@
1
1
1
. lmn . ijk Fmj Fnk
. lmn . IJK x m, J x n, K
2J
2J
1
1
. lmn . IJK Fli Fmj Fnk
. lmn . IJK x l , I x m, J x n, K
6
6
li
il
det ( F )
(2.59)
(2.60)
dJ
dx p ,Q
which becomes:
wx
wx
wx
1
. lmn . IJK l , I x m, J x n, K xl , I m, J x n, K xl , I x m, J n, K
wx p ,Q
6
wx p ,Q
wx p ,Q
(2.61)
2 CONTINUUM KINEMATICS
dJ
dx p ,Q
167
>
1
. lmn . IJK E lp E IQ x m, J x n, K E mp E JQ xl , I x n, K E np E KQ x l , I x m, J
6
>
>
1
. pmn . QJK x m, J x n, K . lpn . IQK xl , I x n, K . lmp . IJQ xl , I x m, J
6
1
. pmn . QJK x m, J x n, K . pnl . QKI x n, K xl , I . plm . QIJ xl , I x m, J
6
1
1
. pmn . QJK x m, J x n, K
. pmn . qjk Fmj Fnk
2
2
@
(2.62)
Referring to the definitions for cofactors and inverting tensors, the following relationship
holds:
dJ
dF pq
dJ
dx p ,Q
1
. pmn . qjk Fmj Fnk
2
>cof (F )@ pq F Fqp1
(2.63)
JX Q , p
We could have obtained the above equation by means of the definition for third invariant
derivatives in respect to tensors, (see Chapter 1), i.e.:
w> III F @
wF
w>det F @
wF
III F F T
J F T
(2.64)
>cof (F )@
1
. pmn . qjk Fmj1 Fnk1
2
&
The derivative of the equation (2.59) with respect to X becomes:
J 1 x q , P
w JX Q , p
wX Q
1
(2.65)
pq
{ ( JFqp1 ) ,q
; X&
0p
JF T
&
0
(2.66)
or
w J 1 x q , p
wx q
{ ( J 1 Fqp ) , q
0p
J 1 F T
; x&
&
0
(2.67)
JX
q , p ,q
,q
>
1
. pmn . qjk Fmj ,q Fnk Fmj Fnk ,q
2
1
. pmn . qjk x m, jq x n,k x m, j x n,kq
2
. jkq
@
(2.68)
0p
0 km .
0p .
& &
&
Using the above definitions, it can be shown that, if u( x , t ) and ( x , t ) represent a vector
and a second-order tensor field, respectively, they satisfy the following relationships:
>
>
@
@
& &
& &
X& u( X , t ) J x& J 1F u( x , t )
&
&
X& ( X , t ) J x& J 1F ( x, t )
(2.69)
168
>J 1 F u( x, t )@
& &
J x&
J J 1 Fij u j
,i
indicial
o J J 1 Fij u j
&
wxi wu j ( x , t )
wX j
wxi
Fij u j ,i
J J 1 Fij ,i u j J 1 Fij u j ,i
,i
&
wu j ( x , t ) wxi
wxi
wX j
0j
&
wu j ( X , t )
(2.70)
& &
X& u( X , t )
wX j
Likewise:
J x&
>J 1 F ( x, t )@
&
J J 1 Fik V kj ,i
J x&
Fik V kj ,i
>J 1 F ( x, t )@
&
&
wxi wV kj ( x , t )
wX k
wxi
&
X& ( X , t )
J J 1 Fik ,i V kj J 1 Fik V kj ,i
0k
&
&
wV kj ( x , t ) wxi
wV kj ( X , t )
wxi
wX j
wX j
indicial
o J J 1 Fik V kj
,i
(2.71)
&
Problem 2.5: Let G ( X , t ) be a scalar field in Lagrangian (material) description. Find the
&
&
relationship between the material gradient of G ( X , t ) , i.e. X& G( X , t ) , and the spatial
&
&
gradient of G ( x , t ) , i.e. x& G( x , t ) .
&
Solution: Remember that a Lagrangian variable G ( X , t ) can be expressed in the Eulerian
(current) configuration by means of the equations of motion, i.e.:
&
& &
&
G ( X , t ) G( X ( x , t ), t ) G( x , t ) .
Then, from the scalar gradient definition we obtain:
&
& &
&
&
&
&
wG( X , t ) wG ( X ( x , t ), t ) wx wG ( x , t )
X& G ( X , t )
&
&
&
& F x& G ( x , t ) F
wx
wX
&
&
wx
2.5.4
wx
wX
wX
wx
&
wG ( X , t )
&
F 1
wX
&
X& G( X , t ) F 1
&
&
xi ( X , t ) X i
and
&
u i ( x, t )
&
xi X i ( x, t )
(2.72)
&
&
Taking the partial derivative of u i ( X , t ) with respect to the material coordinates X , we
obtain:
Indicial notation
&
wu i ( X , t )
wX j
&
ui, J ( X , t )
&
wxi ( X , t ) wX i
wX j
wX j
Fij E ij
Tensorial notation
(2.73)
&
& &
X& u( X , t ) { J ( X , t )
F 1
2 CONTINUUM KINEMATICS
169
&
Taking now the partial derivative of equation (2.72) with respect to spatial coordinates x ,
we obtain:
Indicial notation
Tensorial notation
&
&
wu i ( x , t ) wxi wX i ( x , t )
wx j
wx j
wx j
&
1
u i , j ( x , t ) E ij Fij
(2.74)
& &
&
x& u( x , t ) { j ( x , t ) 1 F 1
J ij
j ij
&
wu i ( X , t )
wX j
wu1
wX 1
wu 2
wX
1
wu 3
wX 1
wu1
wX 2
wu 2
wX 2
wu 3
wX 2
&
wu i ( x , t )
wx j
wu1
wx1
wu 2
wx
1
wu 3
wx1
wu1
wx 2
wu 2
wx 2
wu 3
wx 2
wu1
wX 3
wu 2
wX 3
wu 3
wX 3
wu1
wx3
wu 2
wx3
wu 3
wx3
wx1
wx1
wx1
1
wX 2
wX 3
wX 1
wx 2
wx 2
wx 2
1
wX
wX 2
wX 3
1
x
x
x
w
w
w
3
3
3
1
wX 1
wX 2
wX 3
wX 1
1
wx1
wX 2
wx
1
wX 3
wx1
j ( x , t ) 1 F 1
&
j ( x, t ) F
wX 1
wx 2
wX 2
1
wx 2
wX 3
wx 2
1 F F
1
(2.75)
wX 1
wx3
wX 2
wx3
wX 3
1
wx 3
(2.76)
F 1
(2.77)
and by comparing the above equation with that in (2.73) we can draw the conclusion that
&
&
J ( X , t ) and j ( x , t ) are interrelated by:
&
J ( X , t)
&
j ( x, t ) F
& &
& &
X& u( X , t ) x& u( x , t ) F
(2.78)
u 2
u
3
2 X 12 X 1 X 2
X 22
0
By definition, a material curve is always formed by the same particles. Let OP and OT be
material lines in the reference configuration, where O( X 1 0, X 2 0, X 3 0) ,
P ( X 1 1, X 2 1, X 3 0) and T ( X 1 1, X 2 0, X 3 0) . Find the material curves in the
current configuration. Also find the deformation gradient.
Solution:
a) The equations of motion can be obtained by means of the displacement field, (see Eq.
(2.72)), i.e.:
ui
xi X i
170
x1
substituti ng
o x 2
the values of u1 ,u 2 ,u 3
x
3
x1 u1 X 1
x2 u 2 X 2
x u X
3
3
3
X 1 2 X 12 X 1 X 2
X 2 X 22
X3
Then, to obtain the material curve, one need only substitute the material coordinates with
the particles belonging to the line OP in the equations of motion, (see Figure 2.14). Notice
that the material curve OP in the current configuration is no longer a straight line, but the
line OT is still a straight line in the current configuration (see Figure 2.15).
The components of the deformation gradient, (see Eq. (2.55)), can be obtained as follows:
F jk
wx1
wX 1
wx 2
wX
1
wx 3
wX 1
wx1
wX 2
wx 2
wX 2
wx 3
wX 2
wx1
wX 3
wx 2
wX 3
wx 3
wX 3
X1
(1 4 X 1 X 2 )
0
1
2X 2
0
0
0
0
1
2.5
material curve
x2
1.5
1
0.5
Current Conf.
Reference Conf.
O0
0.5
1.5
2.5
3.5
x1
4.5
2 CONTINUUM KINEMATICS
171
Reference Conf.
x2
0.1
0.08
Reference Conf.
0.06
0.04
0.02
0
O0
0.5
1.5
2.5
3.5
x2
Current Conf.
0.1
0.08
0.06
0.04
0.02
0
Current Conf.
O0
0.5
1.5
2.5
3.5
x1
2.5.5
2.5.5.1
&
w wxi ( X , t )
wt wX j
&
w wxi ( X , t )
wX j
wt
x i , J
&
wvi ( X , t )
wX j
xi
(2.79)
vI ,J
& &
Expressing velocity in Eulerian coordinates, i.e. vi ( X ( x , t ), t ) , and by using the chain rule of
& &
wvi ( X ( x , t ), t )
wX j
vI ,J
&
&
wvi ( x , t ) wx k ( X , t )
wx k
wX j
vi , k x k , J
wx k
wx k
vi , k
l ik
wX j
wX j
(2.80)
l ik Fkj
(2.81)
172
where
&
& &
( x , t ) x& v ( x , t )
F F 1
m
m s
(2.82)
&
Problem 2.7: Let dx be a differential line element in the current configuration. Find the
&
material time derivative of dx .
Solution:
D &
dx
Dt
&
&
D
D
( F ) dX F
( dX )
Dt
Dt
&
D
( F dX )
Dt
&
&
&
&
X
d
&
0
&
dx
dx { x& v dx
Dt i
2.5.5.2
v i , k dx k
&
wv i ( x , t )
dx k
wx k
sym
l
skew
1
1
(l l T ) (l l T ) D W
2
2
(2.83)
sym
1 wvi wv j
2 wx j wxi
; Wij
1 wv i wv j
2 wx j wxi
(2.84)
The spin tensor has three independent components and can be represented as:
Wij
0
W
21
W31
W12
0
W32
W13
W23
0
W12
0
0
W
12
W13
W23
W13
W23
0
0
w
3
w2
w3
0
w1
w2
w1
0
(2.85)
where wi are the axial vector components associated with the antisymmetric tensor W .
&
&
We can also define the vorticity vector field as 2 w . Moreover, as we saw in the chapter on
tensors: given an antisymmetric tensor, the following holds:
1
. kij Wij
(2.86)
2
&
&
&
&
&
In Chapter 1 we proved that 2 w rot (v ) x& v where w is the axial vector associated
&
with the antisymmetric tensor ( x& v ) skew . Therefore, the vorticity vector can be expressed
Wij
as:
wk . kij
or
wk
2 CONTINUUM KINEMATICS
&
&
&
&
&
2w rot (v ) x& v
173
Vorticity vector
(2.87)
& &
wv
(2.88)
When D 0 motion is characterized by rigid body motion, i.e. the distance between
particles does not change. Furthermore, the condition
proved by:
&
&
D ( dx )
l dx
Dt
&
(D W ) dx
&
D ( dx )
Dt
&
&
w dx is satisfied which is
& &
&
W
d
x
w
dx
Antisymmetric
tensor property
(2.89)
To prove that D 0 characterized by rigid body motion our starting point is the definition
of rigid body motion in which the distance between particles does not change, hence the
&
& 2
magnitude of dx does not change over time. Taking the material time derivative of dx
we obtain:
&
& D &
D &
D & &
( dx dx )
( dx ) dx dx
( dx )
Dt
Dt
Dt
& D &
2 dx
( dx ) (see Problem 2.7)
Dt
(2.90)
&
&
&
&
2 dx l dx 2 dx (D W ) dx
&
&
&
&
&
&
&
&
2 dx D dx 2 dx W dx 2 dx D dx 2 W : ( dx dx )
&
&
2 dx D dx
&
&
where we have used the property A skew : B sym 0 W : ( dx dx ) 0 . So, according to
&
(2.90), the magnitude of dx does not change over time if D 0 .
D &
dx
Dt
If the spin tensor is a zero tensor, W 0 , the velocity field is said to be irrotational, thus
&
& &
x& v 0 . In Problem 2.3 the following relationship was validated
v2 1 &
& &
& &
x& v v x& x& v v which can contrast with:
2
2
& &
&
&
&
&
&
&
&
&
x v v D W v D v W v D v W v (W v W v )
>D W @ v& 2W v& 1 l l T l l T v& 2W v&
(2.91)
2
&
& 1
&T & &
& &
1
T
2 l v 2W v
2 x& v v x& v v
2
2
&
&
The term 2 x& v T v can be written in indicial notation as 2v j ,i v j , which is equivalent to
&2
& &
( v ), i (v 2 ), i (v v ), i (v j v j ), i v j ,i v j v j v j ,i 2v j v j ,i , Thus:
>
>
>
v2 &
& &
& &
x& v v x& x& v v
2
(2.92)
174
2.5.5.3
The material time derivative of the spatial deformation gradient ( F 1 ) is obtained from the
material time derivative of F 1 F 1 , i.e.:
D
D
( F F 1 )
1
Dt
Dt
1
DF
0
F 1 F DF
Dt
Dt
F F 1 F F 1 0
(2.93)
Therefore:
F F 1
F F 1
1
F
F F 1
F 1
F 1 F
(2.94)
F 1
Indicial notation
F 1 l
Fij1
(2.95)
Fik1 l kj
NOTE: In this publication we adopt the following notation to represent the material time
.
DF 1
{ F 1 { F 1 .
derivative of the inverse of a tensor:
Dt
2.5.5.4
The material time derivative of the Jacobian determinant can be evaluated by the definition
of the second-order tensor determinant:
J
wxi
wX j
D( J )
{J
Dt
. PQR
wx1 wx 2 wx3
wX P wX Q wX R
(2.96)
According to equations (2.79) and (2.80) the following relationships are valid:
x1, P {
&
wx1 ( X , t )
wX p
&
wv1 ( X , t )
wX p
&
&
wv1 ( x , t ) wx s ( X , t )
wx s
wX P
x 2,Q
wv 2 wx s
wx s wX Q
v 2, s x s ,Q ; x 3, R
wv 3 wx s
wx s wX R
v1, s x s , P
(2.97)
v 3, s x s , R
(2.98)
The first term on the right hand side of the equation in (2.98) can be expressed as:
v1,1 . PQR x1, P x 2,Q x 3, R v1, 2 . PQR x 2, P x 2,Q x 3, R v1,3 . PQR x3, P x 2,Q x3, R
v1,1 J
(2.99)
2 CONTINUUM KINEMATICS
175
in which the following was validated: . PQR x 2, P x 2,Q x3, R . PQR x3, P x 2,Q x3, R 0 , since these
relationships represent a matrix determinant that has two equal rows (linearly dependent).
Similarly, we obtain: . PQR x1, P v 2, s x s ,Q x3, R v 2, 2 J and . PQR x1, P x 2,Q v3, s x s , R v 3,3 J after which
the equation in (2.98) can be rewritten as:
J
v1,1 J v 2, 2 J v3,3 J
(2.100)
v k ,k J
&
F x& v
&
J x& v
& Material time derivative of the
J Tr ( x& v ) Jacobian determinant
J Tr ( l )
J Tr (D)
(2.101)
where we have used the equation in which the trace of an antisymmetric tensor is zero,
Tr ( l ) Tr (D W ) Tr (D) Tr (W ) Tr (D) .
The material time derivative of the Jacobian determinant could also have been obtained as,
(see Chapter 1):
D III F
Dt
D III F DF
:
DF
Dt
( III F F T ) : ( l F )
III F l ik E ik
III F l kk
(2.102)
DFij
D
>det F @
Dt
by:
D wx i ( X , t )
D wx i ( X , t )
cof Fij
cof Fij
Dt wX j
wX j
Dt
&
and considering that v i ( x ( X , t ), t ) , we can state that:
D
>det F @ wvi wx k cof Fij
Dt
wx k wX j
D
>det ( F )@
Dt
D
v i cof Fij
wX j
wv i
det Fij
wx i
T
det Fij
Jv i ,i
wv i
Fkj F ji
wx k
F
ij
1
1
det Fij
wv i
E ki det Fij
wx k
176
quantity
and
&
dx
(ds) 2
&
dx
Reference
configuration - t
X 3 , x3
0(t 0 )
&
dX
& &
dx dx
&
&
F ( X , t ) dX
P
&
X
I 1 , e 1
&
dx
ds
&
dx
Pc
Current
&
x
I 2 , e 2
dS
Qc
configuration - t
I 3 , e 3
O
&
dX
PcQc
&
dX
PQ
&
dx
Q
B0
(2.104)
X 2 , x2
X 1 , x1
t1
2 CONTINUUM KINEMATICS
2.6.1
177
The relationship (ds ) 2 (dS ) 2 can be expressed in the material description as:
(ds ) 2 (dS ) 2
& &
& &
dx dx dX dX
&
& &
F dX F dX dX dX
&
&
& &
dX F T F dX dX dX
&
&
dX F T F 1 dX
&
&
dX C 1 dX
&
&
dX 2 E dX
(ds ) 2 (dS ) 2
dx i dx i dX k dX k
Fik dX k Fij dX j E kj dX k dX
(2.105)
where C is known as the right Cauchy-Green deformation tensor, also known as the Green
deformation tensor, and is defined as:
&
C ( X , t)
FT F
FT F
F 1 F T
(2.107)
We can now introduce the left Cauchy-Green deformation tensor (b) , also known as the Finger
deformation tensor, which we can find in the spatial configuration, and is defined as:
&
b( x, t )
F FT
(2.108)
NOTE: The word right is always associated with material configuration, meanwhile left is
related to spatial configuration.
The inverse of b , which is also in the current configuration, is given by:
&
b 1 ( x , t )
F T F 1
(2.109)
F 1 F T
C 1 the
inverse
o B 1
FT F
(2.110)
In the subsection Polar Decomposition more details will be provided about the configurations
in which these tensors appear.
We can now present some relationships and properties of C and b :
F 1 b F
F C F 1
F C F
; C 1
; b 1
F F
F 1 b 1 F
F C 1 F 1
F F
bb b
(2.111)
2
J2
det (C )
det (C )
(2.112)
III C .
178
Ib
IC
II b
II C
III b
III C
>
1 2
I C Tr (C 2 )
2
1
det (C )
. ijk . pqr C ip C jq C kr
6
The relation I b
Tr (C )
Tr ( F
Tr (C )
(2.113)
J
1
3
1
3
TrC 3 TrC 3 TrC TrC
3
2
2
F)
Tr ( F F T )
Tr (b) .
Furthermore,
the
relationship
Tr (b ) is also valid.
III b
III C
>det(F )@2
III b
The tensors C and b are positive definite symmetric tensors which was proven in
Problem 1.25 in Chapter 1.
Returning to equation (2.105), we now introduce the Green-Lagrange strain tensor denoted by
E , also called the Lagrangian finite strain tensor or the Green-St_Venant strain tensor, and
defined as:
Tensorial notation
&
E( X , t)
The Green-Lagrange
strain tensor
Indicial notation
1 T
( F F 1)
2
1
(C 1)
2
E ij
1
( Fki Fkj E ij )
2
1
(C ij E ij )
2
(2.114)
1 T
( F F 1) T
2
1 T
( F F 1)
2
(2.115)
The Green-Lagrange strain tensor ( E ) can also be expressed in function of the material
& &
displacement gradient tensor, J { X& u( X , t ) . To do this, we start from the equation in (2.73),
i.e.:
F
Fij
J 1
J ij E ij
wu i
E ij
wX j
u i , J E ij
(2.116)
( J 1)
(J
C ij
1)
( J T 1) ( J 1)
J T J J T J 1
Fki Fkj
(u k , I E ki )(u k , J E kj )
u k , I u k , J u k , I E kj E ki u k , J E ki E kj
u k , I u k , J u j , I u i , J E ij
(2.117)
or
C
1 J T J J T J
2E
C ij E ij u k , I u k , J u j , I u i , J
2 Eij
(2.118)
2 CONTINUUM KINEMATICS
179
then:
E
1
J J T J T J
2
E ij
1 wu i wu j wu k wu k
2 wX j wX i wX i wX j
(2.119)
2
2
2
wu
wu
wu1 1 wu1
2 3 ,
wX 1 2 wX 1
wX 1
wX 1
E 22
wu 2 1 wu1
wX 2 2 wX 2
E 33
2
wu
wu 3 1 wu1
2
wX 3 2 wX 3
wX 3
E12
2 wX 2 wX 1 wX 1 wX 2 wX 1 wX 2 wX 1 wX 2
E 21
E13
2 wX 3 wX 1 wX 1 wX 3 wX 1 wX 3 wX 1 wX 3
E 31
E 23
wu wu1 wu 2 wu 2 wu 3 wu 3
1 wu 2 wu 3
1
2 wX 3 wX 2 wX 2 wX 3 wX 2 wX 3 wX 2 wX 3
wu
2
wX 2
wu
3
wX 2
2
wu
3
wX 3
E 32
x2
x3
X1 4X1X 2
X 2 X 22
X 3 X 32
X
X
X 3 (1 4 X )
w
w
w
2
4X1
1
2
wx 2
wx 2
wx k wx 2
0
1
2X 2
Fkj
wX j wX 1 wX 2 wX 3
0
0
wx 3
wx 3
wx 3
wX 1 wX 2 wX 3
(2.120)
And,
0
1 2 X 3
0
180
0
0 (1 4 X 2 )
4X1
(1 4 X 2 )
4X
X
1
2
0
0
1
2X 2
1
2
0
0
1 2 X 3
0
0
(1 4 X 2 ) 2
(1 4 X 2 ) 4 X 1
0
2
2
0
(1 4 X 2 ) 4 X 1 ( 4 X 1 ) (1 2 X 2 )
2
0
0
(1 2 X 3 )
Fki Fkj
0
0
1 2 X 3
(1 4 X 2 ) 4 X 1
0
1
2
2
(1 4 X 2 ) 4 X 1 ( 4 X 1 ) (1 2 X 2 ) 1
0
2
2
0
0
(
1
2
)
1
X
3
E ij
Problem 2.10: Obtain the principal invariants of E in terms of the principal invariants of
C and b .
Solution:
The principal invariants of E are given by:
Tr ( E )
IE
1
Tr (C 1)
Tr ( E )
1
Tr (C 1)
2
where
I E2
2 I C 3
III E
1 2
I C 6I C 9
4
>
1
>Tr (C ) Tr (1)@ 1 I C 3
2
2
>
1 2
I E Tr ( E 2 )
2
1
1
1
Tr (C 1)
Tr (C 1) 2
Tr C 2 2C 1
4
4
2
1
Tr C 2 2 I C 3
4
The term Tr C 2 can be obtained as follows:
Tr ( E 2 )
>
C C
14 >Tr C 2 Tr C Tr 1@
2
C ijc
C1 C 2 C 3 2
C12
0
0
0
C 22
0
0 Tr C 2
2
C3
C12 C 22 C 32
C12 C 22 C 32 2 C1 C 2 C1 C 3 C 2 C 3
II C
C12 C 22 C 32
Therefore we have:
Tr ( E 2 )
det ( E )
1
(C 1) , the principal invariants can also be expressed as:
2
IE
>
1 2
I E Tr ( E 2 )
2
II E
1 2
I C 2 II C 2 I C 3
4
I C2 2 II C
2 CONTINUUM KINEMATICS
II E
1 1 2
1
I C 6 I C 9 I C2 2 II C 2 I C 3
2 4
4
181
1
2 I C II C 3
4
det ( E ) det (C 1)
2
det (C 1)
0
0
1
det >(C 1)@
2
C1 1C 2 1C 3 1
C3 1
C1C 2 C 3 C1 C 2 C1C 3 C 2 C 3 C1 C 2 C 3 1
III C II C I C 1
Then:
III E
1
III C II C I C 1
8
In short we have:
IE
II E
III E
2.6.2
1
I C 3
2
1
2 I C II C 3
4
1
III C II C I C 1
8
IC
2I E 3
II C
4 II E 4 I E 3
III C
8 III E 4 II E 2 I E 1
In the previous subsection, (ds ) 2 (dS ) 2 was expressed in the material description.
Alternatively, it can be expressed in spatial description, i.e.:
(ds) 2 (dS ) 2
& &
& &
dx dx dX dX
& &
&
&
dx dx F 1 dx F 1 dx
& &
& T
&
dx dx dx F F 1 dx
&
&
dx 1 F T F 1 dx
&
&
dx 1 c dx
&
&
dx 2e dx
( ds ) 2 ( dS ) 2
dx i dx i dX k dX k
E ij dx i dx j X k ,i dx i X k , j dx j
(E ij X k ,i X k , j ) dx i dx j
(E ij c ij ) dx i dx j
(2.121)
2eij dx i dx j
where we have introduced c known as the Cauchy deformation tensor, and defined as:
&
c( x, t )
F T F 1
(2.122)
&
1
(1 c )
2
1
(1 F T F 1 )
2
1
(1 b 1 ) The Almansi strain tensor
2
(2.123)
182
eij
1
(E ij Fki1 Fkj1 )
2
1
(E ij cij )
2
(2.124)
c b 1
F F T F T F 1
C B 1
F T F F 1 F T
(2.125)
The Almansi strain tensor ( e ) can also be expressed in terms of the spatial displacement
& &
gradient tensor, x& u( x , t ) { j 1 F 1 :
e
>
@ >
1
1
1
1 F T F 1
1 (1 j ) T (1 j )
1 (1 j T ) (1 j )
2
2
2
1
1
1 1 j jT jT j
j jT jT j
2
2
>
(2.126)
1
j jT jT j
2
eij
Indicial notation
1 wu i wu j wu k wu k
2 wx j wxi
wxi wx j
(2.127)
Both tensors E and e are symmetric second-order tensors and the relationship between
them can be obtained starting from the definition in (2.123), 2e (1 F T F 1 ) :
2F T e F
F T (1 F T F 1 ) F
F T 1 F F T F T F 1 F
FT F 1
2E
(2.128)
FT e F
(2.129)
F T E F 1
(2.130)
and:
e
OBS.: In rigid body motion the relation (ds ) 2 (dS ) 2 is zero, so the strain
tensors ( E , e ) must be zero tensors at any time during motion.
Figure 2.17 summarizes some equations by the use of deformation and strain tensors.
2 CONTINUUM KINEMATICS
Reference
configuration
183
B0
Current
configuration
B
&
X
F 1 b F
Reference Conf.
&
C ( X , t) F T F
&
B( X , t ) F 1 F T
&
1
E( X , t)
C 1
2
C 1
C 1
F T e F
F C F 1
b 1
e
Current Conf.
F 1 b 1 F
&
x
&
b ( x , t ) F F T c 1
&
c( x, t ) F T F 1 b 1
&
1
e( x, t )
1 c
2
F C 1 F 1
F T E F 1
1 2 E E
E 1 2 E E C
1 E 2E E
2.6.3
2.6.3.1
The material time derivative of the right Cauchy-Green deformation tensor, C , is obtained
as follows:
D
D
(C ) { C
( F T F ) F T F F T F
Dt
Dt
FT l T F FT l F FT l T l F
2l
2F T D F
2.6.3.2
(2.131)
sym
The material time derivative of the Green-Lagrange strain tensor, E , is obtained by means
of the equation in (2.114), the result of which is:
D
( E ) { E
Dt
D
Dt
1 T
2 ( F F 1)
D
Dt
1
2 (C 1)
1 T
( F F F T F )
2
1
C
2
(2.132)
184
E
1
C
2
F T D F
(2.133)
and after some algebraic work we can obtain the inverse relationship:
1 T
F C F 1
2
D F T E F 1
(2.134)
The equation in (2.133) could have been obtained by means of the equation in (2.90), i.e.:
>
D
(ds ) 2 (dS ) 2
Dt
>
D
(ds ) 2
Dt
D & &
>dx dx @
Dt
D & &
>dx dx @
Dt
>
&
D &
dX 2 E dX
Dt
&
&
&
&
&
&
2d,
X E dX 2dX E dX 2dX E d,
X
&
&
0
(2.135)
&
&
2dX E dX
Then we have:
&
&
2dX E dX
D & &
>dx dx @ 2dx& D dx&
Dt
&
&
2 dX F T D F dX
(2.136)
The material time derivative of the inverse of the right Cauchy-Green deformation tensor
can be obtained by considering that if C 1 C 1 , it follows that:
D
(C 1 C )
Dt
D
(1) C 1 C C 1 C 0 C 1 C
Dt
C 1 C 1 C C 1
Also by referring to C 1
F 1 F T and C
C 1
C 1 C
(2.137)
2 F 1 D F T
(2.138)
of
the
Left
Cauchy-Green
The material time derivative of the left Cauchy-Green deformation tensor ( b ), (see
equation (2.108)), is given by:
D
(b) { b
Dt
D
(F F T )
Dt
F F T F F T
F FT
F FT l
b b l T (2.139)
D
( F T F 1 ) F T F 1 F T F 1
Dt
b 1
l
b 1 b 1 l (2.140)
So, we can see the tensors b and b 1 are still symmetric tensors.
The material time derivative of the Piola deformation tensor B is given by:
2 CONTINUUM KINEMATICS
D
( B ) { B
Dt
2.6.3.5
185
D
( F 1 F T ) F 1 2D F T
Dt
(2.141)
The material time derivative of the Almansi strain tensor, e , can be obtained by means of
equation (2.123), the outcome of which is:
D 1
(1 F T F 1 )
Dt 2
1 T
1
T
( F F F F 1 )
2
D
(e ) { e
Dt
D 1
(1 c )
Dt 2
1
(c)
2
(2.142)
F T e F F T e F F T e F
F T D F
(2.143)
F , we obtain:
F T e F F T e F F T e F
F e F F 1 F T F T e F F 1 F T F T e F F 1
F T F T e e e F F 1
T
F T l F e e e l F F 1
T
D F
D
D
(2.144)
Thus,
D e e l l
(2.145)
Problem 2.12: Obtain the material time derivative of the Jacobian determinant ( J ) in
terms of ( E ), ( C ), ( F ).
Solution:
This was obtained in (2.101) when J J Tr (D ) , where D is the rate-of-deformation tensor
which is related to E by means of the relationship D F T E F 1 , (see equation
(2.134)), then:
J
J Tr (D)
J Tr F T E F 1
J Fki1 E kp F pj1E ij
J Fki1 F pi1 E kp
J F
T
J ( F 1 F T ) : E
E F 1 : 1
J C 1 : E
J 1
C :C
2
The J can still be expressed in terms of F . To this end let us consider the following
J
1
Fsk Fsp Fsk Fsp , (see Eq. (2.132)). Then, J can also be expressed by:
2
J 1 1
1
J Fki1 F pi1 E kp J Fki1 F pi1 Fsk Fsp Fsk Fsp
Fki F pi Fsk Fsp Fki1 F pi1 Fsk Fsp
2
2
J
J 1
E si Fki1 Fsk E si F pi1 Fsp
Fks Fsk F ps1 Fsp JFts1 Fst JFst Fts1
2
2
JF T : F JF : F T
equation E kp
In short, there are various different ways to express the material time derivative of the
Jacobian determinant:
186
J
J C 1 : E
J Tr (D )
J Tr (C 1 E )
J 1
C :C
2
J
Tr (C 1 C )
2
JF : F T
J Tr ( F F 1 )
2.6.4
&
dx
Current
configuration - t
&
&
F ( X , t ) dX
X 3 , x3
B0
Rc
Q
& 4
dX ( 2) dX& (1)
P
&
X
T & Qc
&
dx ( 2 )
dx (1)
Bt
Pc
&
x
PQ
I 3 , e 3
PR
I 1 , e 1
I 2 , e 2
X 2 , x2
&
dX (1)
&
dX ( 2)
dS (1)
dS ( 2)
PcQc
&
dx (1)
ds (1)
&
dx ( 2)
ds ( 2)
PcRc
X 1 , x1
&
&
dx ( 2 )
dX (1)
&
(dx (1) ) j
F jk
&
dX k(1)
(2.146)
&
dX ( 2 )
&
( dx ( 2 ) ) j
F jk
&
dX k( 2 )
(2.147)
&
and
where F jk
&
2 CONTINUUM KINEMATICS
&
&
F dX (1) F dX ( 2 )
&
&
T
dX (1) F
F dX ( 2 )
&
&
dx (1) dx ( 2)
dx k(1) dx k( 2 )
187
&
&
dX (1) C dX ( 2)
C N
dS (1) dS ( 2 ) M
(2.148)
dX i(1) C ij dX (j 2)
dS (1) dS ( 2 ) M i C ij N j
(1 2 E ) N
dS (1) dS ( 2 ) M
&
dS (1) dS ( 2 ) M i (E ij 2 E ij ) N j
&
&
&
F 1 dx (1) F 1 dx ( 2)
&
& ( 2)
T
1
dx (1) F
F
dx
dX k(1) dX k( 2 )
&
&
dx (1) c dx ( 2 )
(1)
ds ds
( 2)
m c n
(1)
( 2)
m (1 2e ) n
ds ds
1
(2.149)
&
We use the equation in (2.148) to evaluate the magnitude of dx (1) and dx ( 2) in terms of the
&
&
deformation tensors. To do this, in equation (2.148) we enforce that dx ( 2) dx (1) , which
leads to:
&
&
dx (1) dx (1)
&
dx (1)
&
&
dx ( 2 ) dx ( 2 )
&
dx ( 2 )
C M
& (1)
dS (1) dS (1) M
dx
(1)
(1)
dS dS M (1 2 E ) M
&
Similarly, we can obtain the magnitude of dx ( 2) as:
2
& ( 2)
dS ( 2 ) dS ( 2) N C N
dx
(2)
( 2)
dS dS N (1 2 E ) N
C M
dS (1) M
(2.150)
dS ( 2 ) N C N
(2.151)
&
&
Now by using the definition in (2.149) we can express the magnitude of dX (1) and dX ( 2)
as:
&
&
dX (1) dX (1)
&
dX (1)
&
dX (1)
ds (1) ds (1) m (1 2e ) m
ds (1) ds (1) m c m
ds (1) m c m
(2.152)
ds ( 2 ) n c n
(2.153)
and
&
&
dX ( 2 ) dX ( 2 )
2.6.4.1
&
dX ( 2 )
&
dX ( 2 )
ds ( 2 ) ds ( 2 ) n (1 2e ) n
ds ( 2 ) ds ( 2 ) n c n
Next we can establish the relationship between the stretch, unit extension and strain
tensors. To do so we can start by defining the stretch (see equation (2.44)). Then the
stretch along direction M , (see Figure 2.18), can be obtained by means of Lagrangian
variables as:
188
&
dx (1)
&
dX (1)
O M
dS (1) M C M
dS (1)
M C M
(2.154)
M (1 2 E ) M
EM
1 2M
&
where we have used the term dx (1) given in (2.150). If we now use the Eulerian variable,
the stretch along direction m , (see Figure 2.18), is defined as:
&
dx (1)
&
dX (1)
O m
ds
(1)
ds (1)
m c m
1
m c m
1
m (1 2e ) m
1
1 2m e m
(2.155)
Later on, we will show that O m O M . Once the stretch has been defined in terms of strain
tensors, and bearing in mind that the unit extension and stretch are related by the definition
in (2.46), i.e. F O 1 , we can express the unit extension along direction M in terms of
Lagrangian variables as:
F M
C M
1
M
(1 2 E ) M 1
M
O M 1
EM
1
1 2M
(2.156)
We can also evaluate the unit extension in terms of Eulerian variables as:
F m
1
1
m c m
1
1
m (1 2e ) m
O m 1
1
1
1 2m e m
(2.157)
F M
1 2 M E M
O M 1
(1 2 E ) M
C M
EM
1
1 2 M
and
O m
F
m
1
m c m
1 2 m e m
O m 1
1
1
1 2 m e m
1
m b 1 m
1
1
m c m
(2.159)
Notice that, for any given motion, if there is no stretch ( O m 1 ) in a particular direction
( m ), it holds that m c m 1 or m e m 0 .
2.6.4.2
Change of Angle
&
&
The angle between the vectors dx (1) and dx ( 2) , (see Figure 2.18), can be obtained by means
&
&
&
&
dx (1)
dx ( 2) cos T , the outcome of
of the definition of the scalar product dx (1) dx ( 2)
which is:
2 CONTINUUM KINEMATICS
cos T
dS
&
&
dx (1) dx ( 2)
& (1)
&
dx
dx ( 2)
dS
M C N
C M N C N
M
dS ( 2) M C N
C M dS ( 2) N C N
M
(1)
(1)
189
C N
M
O M O N
(2.160)
where we have used the equations in (2.148), (2.150) and (2.151). We can summarize the
different ways of expressing cos T as:
C N
M
C M N C N
M
cos T
M C N
O M O N
1 2 E N
M
1 2 E M
M
N 1 2 E N
N 2 M E N
M
1 2 M E M 1 2 N E N
(2.161)
&
&
dX (1) dX ( 2)
& (1)
&
dX
dX ( 2 )
ds
(1)
ds (1) ds ( 2) m c n
m c m ds ( 2) n c n
m c n
m c n
O m O n
m c m n c n
m 1 2e n
m 1 2e m n 1 2e n
(2.162)
where we have used the equations in (2.149), (2.152) and (2.153). Then, we can summarize
cos 4 as:
m c n
O m O n (m c n )
m c m n c n
m 1 2e n
m n 2m e n
m 1 2e m n 1 2e n
1 2m e m 1 2n e n
cos 4
1 2 E N
M
O M O N
(2.163)
N 2 M
E N
M
O M O N
cos 4 2 M E N
O M O N
(2.164)
and
cos 4
(2.165)
2.6.4.3
2.6.4.3.1
Let us consider the Cartesian components of the right Cauchy-Green deformation tensor
at the material point P (particle), (see Figure 2.18).
190
e 3
X3
C ij
C11
C
12
C13
C ji
C13
C 23
C 33
C12
C 22
C 23
C 33
C13
C 23
C 23
C13
C 22
C12
C12
C11
e 1
e 2
X2
X1
M i C ij M j
>1
C11
0 0@ C12
C13
C13 1
C 23 0 C11
C 33 0
C12
C 22
C 23
(2.166)
O X1
M C M
C11
1 2 E11
O X1 ! 0
(2.167)
As we can see, C11 is the stretch measurement along the X 1 -axis. Similarly, C 22 and C 33
show the stretch along X 2 and X 3 , respectively, i.e.:
O X1
C11
1 2 E11
O X2
C 22
1 2 E 22
O X3
C 33
1 2 E 33
1 2
O X1 1
2
1 2
O X2 1
2
1 2
O X3 1
2
E11
E 22
E 33
(2.168)
Therefore, the conclusion is that the diagonal terms E and C are related to the stretches.
Notice that, C is a symmetric positive definite tensor, and if we are working in the C
principal space, it follows that:
O21
C ijc
0
0
0
O22
0
0
2
O3
Eijc
1 2
2 O1 1
1 2
O2 1
2
0
1 2
O 3 1
2
(2.169)
where O1 ! 0 , O 2 ! 0 , O 3 ! 0 are the principal stretches, which by definition are positive real
numbers, (see equation (2.44)). Then, the spectral representations of C and E are
expressed as follows:
2 CONTINUUM KINEMATICS
(a) N
(a)
N
(a) N
(a)
1) N
2
a
a 1
2 (O
2
a
191
a 1
In addition, by means of the spectral representation of C we can define a new tensor such
that U 2 C , where U denotes the right stretch tensor, and where the only possible solution
for U is U C . Since the stretches are by definition positive real numbers, it follows
that the tensor U is definite positive. We can then define the right stretch tensor as:
&
U( X , t )
(a) N
(a)
N
a 1
2.6.4.3.2
e 1 and N
>1
M i C ij N j
C N becomes:
e 2 . So, the product M
C11
0 0@ C12
C13
C12
C 22
C 23
C13 0
C 23 1 C12
C 33 0
(2.173)
With the following we can verify that the above term is related to cos T , (see equation
(2.161)):
cos T
C N
M
C M
N C N
M
M C N
O M O N
C12
1 2 E12
C11 C 22
1 2 E11 1 2 E 22
(2.174)
So, C12 measures the angle change between two differential line elements in the reference
configuration. Therefore, the off-diagonal terms E and C contain information about the
angle change.
Homogeneous Deformation
If we consider an example in which the motion of all the particles is characterized by the
&
same deformation gradient, it follows that F is independent of the position vector X , and
it is therefore only dependent on time, F F (t ) . This type of motion is an example of
&
&
homogeneous deformation. By integrating the equation dx F (t ) dX , we obtain:
& &
(2.175)
F (t ) X c (t )
&
where the constant of integration c shows translational motion, which is only dependent
&
x
on time.
Motion characterized by homogeneous deformation has the following characteristics:
192
2.7.2
We can state that a body undergoes rigid body motion when the distance between particles
are constant during motion. Under these conditions we can conclude that rigid body
motion is a specific case of homogenous deformation. Let us consider a vector in the
&
&
reference configuration A . After motion, this vector is represented by a (deformed
&
&
configuration). Then according to equation (2.175), it follows that a F A . Moreover, as
&
A . In this situation, we
&
FT F
F FT
Q QT
(2.176)
In addition, we find that for rigid body motion the following is satisfied:
1
C 1 0
2
The strain tensors for rigid body motion
1
e
1 b 1 0
2
D 0
E
(2.177)
In rigid body motion the stretches are unitary, since the distance between particles does not
change, and it is possible to check the previous result by means of the spectral
representation of C and E , (see equations (2.170) and (2.171)):
3
2
a
(a) N
(a)
N
a 1
2 (O
2
a
(a )
N
(a)
(2.178)
a 1
(a) N
(a)
1) N
a 1
(a) N
(a )
N
(2.179)
a 1
NOTE: To ensure that the continuum is subjected to rigid body motion, the equation
E 0 or D 0 must be valid for all material points throughout the continuum.
Problem 2.13: Let us consider the following equations of motion:
X1
1
X2
2
1
X1 X 2
2
;
x3 X 3
&
a) Obtain the displacement field ( u ) in the Lagrangian and Eulerian descriptions;
x1
x2
(2.180)
b) Determine the material curve in the current configuration for a material circle defined in
the reference configuration as:
2 CONTINUUM KINEMATICS
X 12 X 22
X3
193
c) Obtain the components of the right Cauchy-Green deformation tensor and the GreenLagrange strain tensor;
d) Obtain the principal stretches.
Solution:
The deformation gradient is given by:
Fij
2 1 0
1
1 2 0
2
0 0 2
wxi
wX j
0.75
2 1 0 X 1
1
1 2 0 X 2
2
0 0 2 X 3
xi
Fij X j
So, we can verify that the proposed example is a case of homogeneous deformation in
& &
which c 0 . The inverse form of the above equation is given by:
4
2
X 1 3 x1 3 x 2
X1
4 2 0 x1
2
4
X 1 2 4 0 x
(2.181)
X 2 x1 x 2
2
2
3
3
3
X 3
0
0 3 x3
X 3 x3
u1 ( X , t ) x1 X 1 X 1 2 X 2 X 1 2 X 2
&
1
1
xi X i u 2 ( X , t ) x 2 X 2
X1 X 2 X 2
X1
2
2
&
u 3 ( X , t ) x3 X 3 0
ui
(2.182)
The components of the Eulerian displacement can be obtained by substituting the Eulerian
description of motion (2.181) into (2.182), the result of which is:
& &
&
1
u1 ( X ( x, t ), t ) 2 X 2 ( x , t )
& &
&
1
X 1 ( x, t )
u 2 ( X ( x , t ), t )
2
&
u ( X ( x& , t ), t ) u ( x& , t ) 0
2
3
&
1 2
4
x1 x 2 u1 ( x , t )
2 3
3
&
1 2
4
x1 x 2 u 2 ( x , t )
2 3
3
(2.183)
The particles belonging to the circle X 12 X 22 2 in the reference configuration will form
a new curve in the current configuration which is defined by:
2
X 12 X 22
2
4
4
2
2 x1 x 2 x1 x 2
3
3
3
3
2 20 x12 32 x1 x 2 20 x 22
18
which is an ellipse equation (Figure 2.20 shows the material curve in different
configurations).
194
F T F and
1
C 1 :
2
C ij
E ij
Fki Fkj
1
C ij G ij
2
2 1 0 2 1
1
1 2 0 1 2
4
0 0 2 0 0
1.25 1 0 1
1
1 1.25 0 0
2
0 1 0
0
C ij
E ij
0
0
2
1.25 1 0
1 1.25 0
0
0 1
0 0 0.125 0.5 0
1 0 0.5 0.125 0
0 1 0
0
0
O21
0
0
0
O22
0
0
O23
C ijc
O 1
0
0
0
O 3
0
O2
0
where O i show the principal stretches. Therefore, to calculate these we need to obtain the
C eigenvalues:
1.25 C
1
1
1.25 C
O21
C ijc
0
0
0
O22
0
0
O23
C1
0 C 2 2.5C 0.5625 0
C 2
O 1 0
0
0
0
2.25
0.25 0
0 O2 0
0 0 O
0
0
1
3
2.25
0.25
1.5 0 0
0 0.5 0
0
0 1
2.0
material curve
1.5
Reference Conf.
Current Conf.
1.0
0.5
x2
0.0
-2
-1
-0.5
-1.0
-1.5
-2.0
x1
2 CONTINUUM KINEMATICS
195
v1
v 2
v
3
3 x1 5 x3
1x1 5 x 2
l ij
wv1
wx1
wv 2
wx
1
wv3
wx1
wv1
wx3
wv 2
wx3
wv3
wx3
wv1
wx 2
wv 2
wx 2
wv 3
wx 2
0 3 1
3
0 5
1 5
0
l :
skew
l ij
Taking into account that l can be decomposed into a symmetric ( l sym { D ) and an
antisymmetric ( l skew { W ) part, i.e. l D W , we can thus conclude that D 0 , which is
a characteristic of rigid body motion.
V R
R
U
,
(2.184)
Right
polar decomposition
R
and det
orthogonality
proper
In the right polar decomposition, we first carry out a transformation just with strain, and
then we make a transformation characterized by a rotation, (see Figure 2.21), whereas, in
the left polar decomposition, we first carry out an orthogonal transformation (rotation) and
then transformation only with strain is applied.
With the right polar decomposition it holds that U R T F , and also by applying the
scalar product between F T and the equation in (2.184) we obtain:
F T
F
C
F T R U
UT U U 2
(2.185)
196
2
a
(a) N
( a ) , (see
N
a 1
(a )
(a)
a 1
stretches, and are positive numbers by definition, and so the tensor U is a positive definite
tensor.
Since the determinant of F is positive, det ( F ) ! 0 , and the determinant of a positive
definite tensor is also positive det (U) ! 0 , we can conclude that R is a proper orthogonal
tensor, i.e. a rotation tensor:
det ( F ) det (R U) det (R ) det (U) det(U) ! 0
B0
&
dX
n (1)
( 3)
N
(2)
N
(1)
N
&
dX
n (3)
Reference
configuration
(2.186)
!0
B0
n ( 2 )
V R
&
dx
&
F dX
R U
&
dx
U
n (3)
&
dx
n (1)
n ( 2 )
Current
configuration
( 3)
N
R
(2)
N
(1)
N
B
FT F
(R U)
U
T
T
(R U)
R U
U U U2
FT F
(V R)T (V R)
RT V V R
RT b R
(2.187)
2 CONTINUUM KINEMATICS
F FT
b c 1
F FT
(V R) (V R)
(R U) (R U) T
V R RT VT
VV
197
R UU RT
(2.188)
R C RT
V2
where c is the Cauchy deformation tensor. And the tensors C and b are interrelated to
each other by:
U RT V R
2.8.1
V R U RT
(2.189)
As we have seen before, the eigenvalues of U represent the principal stretches, O i . Each
(i ) ), i.e.:
principal stretch ( O i ) is associated with a principal direction ( N
for O1
(1)
N
for O 2
(2)
N
for O 3
> N
> N
> N
( 3)
(1)
1
N (21)
N 3(1)
( 2)
1
N (22 )
N 3( 2)
( 3)
1
N (23)
( 3)
N
3
@
@
(2.190)
(a ) N
(a)
N
(2.191)
a 1
U2
3
3
(a) N
( a ) O N
(a) N
( a )
Oa N
a
a 1
a 1
3
3
(a)
(a) (a)
(a)
2 (a)
(a )
OaOa N N
N N
Oa N N
a 1
(2.192)
a 1
Thus,
C
U2
2
a
(a) N
(a)
N
a 1
If we can verify that, U and C are coaxial tensors, then it holds that C U U C U 3 .
Based on the principle that C and b have the same principal invariants, (see equation
(2.113)), then C and b have the same eigenvalues. We can prove this after having defined
the eigenvalues and eigenvectors of U :
O N
UN
a
(2.194)
UN
, O aN
n
(2.195)
N
OaR
,
V n
O a n
(2.196)
198
Thus, we find that U and V have the same eigenvalues O i , but different eigenvectors so
the spectral representation of V and b are given by:
V
n ( a ) n ( a )
2
a
n ( a ) n ( a )
O
a 1
V2
O
a 1
, n . If we
Next, we show F in terms of the eigenvalues of U , O a , and eigenvectors N
R T n n R , it holds that:
consider that U R T F and N
O N
UN
a
O N
RT F N
a
R T O n
RT F N
a
(2.199)
F N
O a n
(2.200)
O a n ( a ) n ( a ) R
a 1
V R
a 1
(a )
n ( a ) n
R
(a )
N
(2.201)
3 1 (a)
R T
n n ( a )
a 1 Oa
R T V 1
O
a 1
(a)
T (a)
R
n
(a)
N
(2.202)
(a)
n ( a ) N
a 1
F 1
1 (a ) (a )
N n
O
a 1
Spectral representation of
the deformation gradient
(2.203)
By means of the spectral representation of the deformation gradient, we can see that F
is neither in the reference nor in the current configuration. It is as if it were straddling both
of them.
By making use of the left polar decomposition, F V R R V 1 F , the spectral
representation of the orthogonal tensor of the polar decomposition can be obtained as
follows:
R
3 1 (a)
3
(a)
(a) (a)
O n n Oa n N
a 1 a
a 1
1
(a)
O a n ( a ) n ( a ) n ( a ) N
O
1 a
Thus:
(2.204)
2 CONTINUUM KINEMATICS
(a)
(a)
N
199
(i )
n (i ) N
a 1
(2.205)
1
C 1 1 U 2 1
2
2
1
1 b 1
2
12 (O
3
2
a
(a ) N
(a)
1) N
a 1
12 1 V 12 (1 O
3
2
2 (a )
a )n
(2.206)
n ( a )
a 1
&
&
&
dx F dX V R dX
&
&
&
dx F dX R U dX
then:
&
V dX
&
R dx
&
dx
&
R
V dX
T
(2.207)
&
dX
V
R dx
1
&
F T
FT
(2.208)
F C F 1
F C F 1
V R C R T V 1
R U C U 1 R T
V C V
R C R
1
UC U
1
F 1 b F
R T V 1 b V R
R T V 1 V 2 V R
R
UU
1
(2.209)
F 1 b F
U 1 R T b R U
U 1 b U
(2.210)
V R
2
RT b R
All the equations obtained above can be appreciated in Figure 2.22. We leave the reader to
make the necessary algebraic operations with the inverse tensors.
200
C
C
R C RT
b R C RT
R
&
dX
V2
b
b 1
V 2
V
&
dX
&
R dX
&
dx
B0
RT b R
C
C 1
R T b 1 R
&
V dX
b
V C * V 1
V b V 1
b 1
V 2
&
dx
&
dX
RT V
B0
V 1 R
FT
F T
Current
configuration
Reference
configuration
FT F
U 1 b U
C 1
V2
b R b RT
U2
&
dx
U 1 b 1 U
&
dx
C 1
U C U 1
R b 1 R T
b 1
R C 1 R T
&
&
dx R dx
&
U dX
b 1
U2
U C 1 U 1
b
b 1
U 2
RT b R
R T b 1 R
2
obtain the tensors U (right stretch tensor), V
Fij
3 3
6 3
2 4
2 CONTINUUM KINEMATICS
F33c
> 0.4082482905;
201
0.4082482905; 0.8164965809@
It is easy to check that the basis formed by these eigenvectors does not form an orthogonal
basis, i.e. m i(1) m i( 2) z 0 , m i(1) m i(3) z 0 , m i( 2) m i(3) z 0 . We can also verify that if D is the matrix
containing the eigenvectors of F :
D
m i(1)
( 2)
m i
m (3)
i
0.6396021491;
0.4264014327
0.6396021491;
0.5570860145; 0.7427813527; 0.3713906764
5 2 2
3 6 2
3 3 4
D 1 0
5 2 2
( F T ) ij
and
D 3 6 2 D 1
10 0 0
0 3 0
0 0 2
3 3 4
The right Cauchy-Green deformation tensor components, C F T F , are given by:
33 31 29
C ij Fki Fkj 31 49 35
29 35 34
9.274739
eigenvecto
r o
c
C 22
3.770098
eigenvector
c
C 33
102.955163
>0.6861511933;
>0.5105143234;
i
(3) > 0.518239;
N
i
(1)
N
i
N ( 2 )
o
eigenvecto
r o
0.7023576528; 0.1894472683@
0.2793856273; 0.8132215099@
0.65470405; 0.550264423@
These eigenvectors constitute an orthogonal basis, so, it holds that AC1 ACT , and
det (AC ) 1 (improper orthogonal tensor):
AC
N (i1)
( 2)
N i
N
(3)
i
0.518239
0.65470405
0.550264423
ACT 0
0
c
C 22
0
0 AC
c
C 33
33 31 29
33 31 29
31 49 35 C ; A 31 49 35 A T
ij
C
C
29 35 34
29 35 34
c
C11
0
0
c
C 22
0
0
c
C 33
0
0
0
In the C principal space we obtain the components of the right stretch tensor, U , as:
c
0 C11
0
0 3.0454455
0
0
O 1 0
c
C 22
0
0
1.9416741
0
U c Ucij 0 O 2 0 0
0 0 O 3 0
c
0
C 33
0
0
10.1466824
U c 1 Ucij1
O1
0
0
1
O2
0
1
O 3
0
0
3.0454455
0
0
1.9416741
0
0
10.1466824
We can evaluate the components of the tensor U in the original space by means of the
transformation law:
202
ACT U cAC
and
ACT U c 1AC
Then, the rotation tensor of the polar decomposition is given by the equation R F U 1 ,
which is a proper orthogonal tensor, i.e. det (R ) 1 .
0.10094326 0.05592536
0.9933191
0.10658955 0.98826538 0.10940847
R ij
9.274739
eigenvecto
r o
c
b22
3.770098
eigenvector
c
b33
102.95516
o
n (i1)
n i( 2 )
eigenvecto
r o
n (i 3)
Note that, the tensors b and C have the same eigenvalues but different eigenvectors. If
the eigenvectors of b constitute an orthogonal basis then it holds that Ab1 AbT , and
det (Ab ) 1 :
Ab
n (i1)
( 2)
n i
n (3)
i
0.238183919
0.6212637156 0.7465251613
0.4898263742 0.1327190337 0.8616587383
0
0 Ab
c
b33
43 37 28
43 37 28
37 49 28 b ; A 37 49 28 A T
b
ij
b
0
28 28 24
28 28 24
0
Since C and b have the same eigenvalues, it follows that Ucij Vijc , i.e.
AbT 0
0
c
b22
c
b11
0
0
c
b22
0
AbT U cAb
0
0
c
b33
Vijc 1 .
Vij
and
AbT V c 1Ab
AbT U c 1Ab
Vij1
The polar decomposition rotation tensor obtained previously has to be the same as the one
obtained by R V 1 F .
2 CONTINUUM KINEMATICS
203
U ij
3
(a) N
( a )
Oa N
ij
a 1
(3)
O 1 N i(1)N (j1) O 2 N i( 2 ) N (j2 ) O 3 N i(3) N
j
Vij
3
O a n ( a ) n ( a )
ij
a 1
R ij
3 (a) (a )
n N
ij
a 1
Fij
3
( a )
O a n ( a ) N
a 1
ij
( 2 ) O n (3) N (3)
O1 n (i1) N (j1) O 2 n (i 2 ) N
j
3 i
j
3
(a) N
(a)
R N
a 1
n ( a ) n ( a ) R
a 1
3
(a ) N
( a )
R O a N
a 1
R U V R
O a n ( a ) n ( a ) R
a 1
As we can verify, the representations of the tensors R and F are not the spectral
representations in the strict sense of the word, i.e., O i are not eigenvalues of F , and
(i ) are eigenvectors of F .
neither n (i ) nor N
2.8.2
Using the right polar decomposition ( F R U ) as seen in (2.184), the material time
derivative of the deformation gradient ( F ) can also be evaluated by:
F
R U R U
(2.211)
R U
l R
l
R U R U
U 1
R U U 1 R U
1
R U
U
R T R U U 1 R T
(2.212)
Thus,
l
U 1 R T
R R T R U
(2.213)
Notice that, in rigid body motion, U 1 U 0 , the spatial velocity gradient becomes
l
R R T . This is a prompt for us to introduce an antisymmetric second-order tensor, the
rate of the material rotation tensor (also called the angular-velocity tensor), and defined as:
8 R R T
8 T
(2.214)
Additionally, the axial vector associated with 8 is called the angular-velocity vector and is
&
denoted by Z .
204
(2.215)
8 T
(R R T )
(R ) R T R
(R T )
Dt
Dt
Dt
R T R R T
R
8
T
R T R 1 R T
R
8 R R 8 8
R T R R T R R T
R
R T 8 8
R
(2.216)
Taking into account (2.213), the rate-of-deformation tensor, D , can also be expressed as:
D
1
(l l T )
2
1
U 1 R T R R T R U
U 1 R T T
R RT R U
2
T
1
T
1
T
T
T
T
T
U R R R
R
R U U
R R R U
(2.217)
>
1
U 1 U 1 U
RT
R U
2
(2.218)
>
1
1
U1 U1 U
RT
( l l T ) R R T R U
2
2
(2.219)
We can now attempt to graphically visualize the tensors obtained above. To do so, let us
consider Figure 2.23, in which the time domain is discretized by means of time increments
't . And, at each time step we represent the right polar decomposition.
2 CONTINUUM KINEMATICS
l U
U
U
't o 0
B ( 2)
't o 0
R
't
.. .
U ( 2)
U (1)
't
lU
205
(1)
't
R ( 2)
't
U 1
U
lR
8 R
R (1)
B0
R
lR
R R T
lU
R l U RT
l
DW
't
B (1)
. ..
B ( 2)
F (1)
't
F ( 2)
F
't o 0
lU
U 1
U
(2.220)
lU
Note that
true that:
lR
lR
lR
R R 1
lR
R R T
lR
(current
(2.221)
206
l
lR
U1 R T
R R T R U
lU
(2.222)
lU
lU
sym
sym
l Rsym
,
0
1
R l U RT R l U RT
2
lU
sym
1
U 1 R T R U
U 1 R T
R U
2
1
U 1 U 1 U
RT
R U
2
>
lU
skew
l Rskew
skew
lU
skew
(2.223)
1
U 1 R T R U
U 1 R T
R U
2
1
U 1 U 1 U
R T R R T
R U
2
>
1
R l U RT R lU RT
2
lR
can also be
R R
T
R R T
can also be
T
(2.224)
Using F
R V R
V
(2.225)
V R
V R V R
V
V
1
V R R
(2.226)
1
we obtain:
(2.227)
1
1
2
1
2
1
2
1
2
1
2
1
2
l
V V
1
V V
1
12 V V
1
V 1 R R T V
V R R T V 1 V 1 V
1 V R R T V 1 V 1 R R T V
V 1 V
2
1
1
V V V 8 V 1 V 1 8 V
2
(2.228)
and
W
l
V V
12 V V
1
V R R T V 1 V 1 R R T V
V 1 V
V V
1
1
V 1 R R T V
V R R T V 1 V 1 V
1 V 8 V 1 V 1 8 V
V 1 V
2
(2.229)
2 CONTINUUM KINEMATICS
207
The tensors obtained above can be appreciated in Figure 2.24, where we represent the rate
of change of R by means of
lR
R R T
B0 (t ) .
R
't o 0
lR
.
.
.
8 R
R ( 2)
B0 ( 2)
R (1)
't
B0
V
lV
V 1
V
lR
V l R V 1
l
V (1)
B0
lV
V (2)
't
R R T
lR
V
't
B0 (1)
't
't o 0
DW
't
B (1)
B ( 2)
. ..
F (1)
't
F ( 2)
't o 0
F
V l R V 1
lR
V R R T V 1
V 8 V 1
Then,
lV
(2.230)
V 1
V
(2.231)
lV
lR
V V 1 V l R V 1
V 1 V R R T V 1
V
(2.232)
208
which is the same equation as that obtained in (2.227). The symmetric part of
obtained as:
D{l
lV
l Rsym
sym
sym
lV
skew
skew
l Rskew
1
1 V 8 V 1 V 1 8 T V
V V 1 V 1 V
2
2
l
can be
(2.233)
is given by:
1
1 V 8 V 1 V 1 8 T V
V V 1 V 1 V
2
2
(2.234)
Let us consider the motion and evolution of the polar decomposition as shown in Figure
2.25. We can introduce a new configuration B0 , which does not change over time. So, we
&
F dX
D &
( dx )
Dt
of change
rate
o
dx
&
F
N Z 0 N
R0 Z 0 R0
(2.235)
R0 N 0
n
of change
rate
o
R N 0
N
lR
of change
rate
o
n l R n
lR
Z n
R Z R
(2.236)
(2.237)
R R R0 R R0
Z R 8 R R0 R Z 0 R0
Z R R0 8 R R0 R Z 0 R0
Z R R0 8 R R Z 0 R0
Z R
8 R R Z0
8 R Z0 RT
(2.238)
Z 0 R0 ,
(2.239)
Starting from (2.239) and referring to the fact that 8 R R T , we can obtain the following
equation for R :
Z R
(R R T ) R R Z 0
R
Z R R Z0
(2.240)
2 CONTINUUM KINEMATICS
209
V
R
lR
lV
B0
n
B0
F
F
N
0
R0
R0
with
F F 1
U 1
U
lR
lR
lR
lR
R R
B0
R
lU
R0 U R0T
l U
U
U
lR
8 R
R0
Z 0 R0
lR
R R R0
R
lR
Z R
N
0
R0 R0T Z 0
R R T Z
R N
n R N
R Z
8 R N
8 R R Z 0 N
Z R N
Z n
R N
is given by:
And, that of N
0
0
R N
n R N
R Z
8 R N
8 R R Z 0 R T n
8 R Z 0 R T n
Z n
(2.241)
210
D
N{N
Dt
R0 N 0 R0 N
R0 N 0 Z 0 R0 N 0 Z 0 N
,0
(2.242)
&
0
( a ) , and the equation in (2.236), N
N
(a)
a 1
a 1
a 1
a 1
a 1
a 1
a 1
Z 0 1 Z 0 N ( a) N (a ) Z 0 N ( a ) N ( a) N ( a) N (a ) Z 0
(2.243)
Similarly,
Z 1 Z n ( a) n (a ) Z n ( a) n ( a) n ( a) n (a ) Z
(2.244)
Z 0 Z 0 ab N ( a) N (b)
Z Z ab n (a ) n (b)
a ,b 1
azb
a ,b 1
a zb
(2.245)
Z ab n ( a ) n (b) R R Z 0 ab N ( a) N (b)
a ,b 1
a ,b 1
a zb
azb
ab
(a)
(b )
N
a ,b 1
azb
0 ab
(a)
(b )
N
a ,b 1
a zb
ab
(2.246)
Z 0 ab n
(a )
(b )
N
a ,b 1
azb
a
b
(
)
(
)
Z ab Z 0 ab n N R T
a ,b 1
a zb
8 R R T
ab
Z 0 ab n ( a ) n (b )
(2.247)
a ,b 1
azb
(a ) N
(a)
N
a 1
a 1
R0
O R
3
O a N (0a ) N 0( a )
a 1
N 0( a ) R0 N 0( a )
O R
a
a 1
R0T
(2.248)
R0 U R0T
where we have introduced the tensor U , which is in the configuration B0 , (see Figure
2.25), and U is given by:
U
O
a 1
(a) N
(a )
N
0
0
of change
rate
o
U
O
a 1
(a) N
(a)
N
0
0
(2.249)
2 CONTINUUM KINEMATICS
211
U
(2.250)
Z 0 U U Z 0 O a N ( a) N (a )
a 1
Z 0 U U Z 0 Z 0 ab (O b O a ) N ( a) N (b)
a ,b 1
a zb
(2.251)
and
3
(2.252)
a ,b 1
a zb
O
(a) N
(a)
N
a 1
(a) N
(b )
Oa )N
0 ab (O b
(2.253)
a ,b 1
azb
R URT
R R0 U R0T R T
(2.254)
R U R T
V
R U R T R U R T R U R T
Z R U R T R U R T R U R T Z T
Z V R U R T
3
V ZT
Z V V Z O a n ( a) n (a )
a 1
or
Z V R O a N (0a ) N (0a) R T
a
U ZT
(2.255)
212
3
O
V
n ( a ) n ( a )
a 1
ab (O b
O a ) n ( a ) n (b )
(2.256)
a ,b 1
azb
U2
U R0T
(2.257)
C
(2.258)
Z 0 C C Z 0 2O a O a N ( a ) N ( a)
a 1
or
3
2O
C
(a ) N
(a)
N
aOa
a 1
2
0 ab (O b
(a ) N
(b )
O2a ) N
a ,b 1
azb
(2.259)
R U R R U R R U R
T 2
V2
R U 2 R T
(2.260)
Z b b Z 2O a O a n ( a) n ( a)
(2.261)
a 1
By fixing
b
2O
n ( a ) n ( a )
aOa
a 1
2
ab (O b
O2a ) n ( a ) n (b )
(2.262)
a ,b 1
a zb
( a ) , becomes:
n ( a ) N
a 1
F
O
a 1
O
3
a 1
3
(a) O
n ( a ) N
a
Z n ( a) N (a ) O a n ( a) Z 0 N ( a )
3
3
( a ) Z O n ( a ) N
( a ) O n ( a ) N
( a ) Z T
O a n ( a ) N
0
a
a
1
a 1
a 1
( a ) O n ( a ) N
( a ) O n ( a ) N
( a )
n ( a ) N
a
a
O
a 1
(a) Z F F ZT
n ( a ) N
0
O
a 1
(a ) Z F F Z
n ( a ) N
0
(2.263)
2 CONTINUUM KINEMATICS
213
Using the same reasoning we made to solve Problem 1.34, we can state that:
ZF
3
3
b 1
a ,b 1
a zb
3
(a )
(b )
( a )
O a n ( a ) N
Z 0 ab N N
a 1
a ,b 1
azb
F Z0
ab O b
(b )
n ( a ) N
(2.264)
a ,b 1
azb
3
0 ab O b
(b )
n ( a ) N
(2.265)
a ,b 1
azb
O
(a)
n ( a ) N
a 1
O Z
(b )
O a Z 0 ab n ( a ) N
ab
a ,b 1
a zb
F F
1
(2.266)
1 (a) (a )
( a )
O a n ( a ) N
N n
a 1
a 1 Oa
1 (b ) (b )
(b )
O b Z ab O a Z 0 ab n ( a ) N
N n
a ,b 1
b 1 Ob
a zb
(2.267)
which becomes:
l
O a
O
a 1
n ( a ) n ( a )
By referring to D F T E F 1
a ,b 1
a zb
ab
Oa
Ob
Z 0 ab n ( a ) n (b)
(2.268)
1 T
F C F 1 , (see equation (2.134)), and using the
2
1 3 1 (a) (a) 3
(a ) N
(a)
n N 2O a O a N
2 a 1 O a
a 1
3 1 (a)
(a )
O N n
a 1 a
1 3 1 (a) (a ) 3
2
2
(a )
(b )
n N
Z 0 ab (O b O a ) N N 1 N (b) n (b)
O
2 a 1 Oa
b 1 b
a ,b 1
azb
(2.269)
which becomes:
D
O a
O
a 1
By referring to W
n ( a ) n ( a )
a ,b 1
azb
0 ab
(O2b O2a ) ( a )
n n (b )
2O a O b
(2.270)
a ,b 1
a zb
ab
Z 0 ab
O2b O2a
2O a O b
(a)
n n (b )
(2.271)
214
&
x
(2.272)
&
&
Find the velocity and the acceleration fields as a function of Z , where Z is the axial vector
T
Q ).
associated with the antisymmetric tensor ( 8 Q
Solution:
&
& &
The material time derivative of x c(t ) Q(t ) X is given by
D & & & &
x { x c Q X
Dt
QT Q
8 Q . The above equation can also be expressed as:
Let us consider that 8 Q
&
& &
v c 8 Q X
& &
& &
v c 8 ( x c )
& & &
&
If 8 is an antisymmetric tensor, it holds that 8 a Z a , where Z (angular velocity vector)
is the axial vector associated with the antisymmetric tensor 8 , (see equation (2.88)). Then,
&
v
&
& &
c 8 ( x c )
& &
& &
c Z ( x c )
&
v
(2.273)
Note that Q(t ) is only dependent on time, hence the axial vector (angular velocity)
& &
associated with 8 is also time-dependent, i.e. Z Z (t ) .
Then, its acceleration is given by:
&
a
&
v
8
, the above equation can also be expressed as:
Q 8 Q
By referring to Q
&
a
&
c& (8
) X
Q 8Q
&
&
c& 8
X
Q X 8Q
&
&
c& 8
Q X 8 8 Q X
c& 8
( x& c& ) 8 8 ( x& c& )
&
& &
&
&
& &
c& Z
( x c) Z >Z ( x c)@
1 w (. ipq Z p x q ) w (. jpq Z p x q )
wx j
wxi
2
(2.274)
& &
where B { Z shows the angular acceleration.
& &
& & &
For a rigid body motion where c 0 , the velocity becomes v Z x whose components
are vi . ipq Z p x q , and the rate-of-deformation tensor D becomes:
D ij
1 wvi wv j
2 wx j wxi
1
. ipq Z p E qj . jpq Z p E qi
2
wx q
wx q
1
. jpq Z p
. ipq Z p
wx j
2
wxi
1
1
. ipj Z p . jpi Z p
. ipj Z p . ipj Z p
2
2
So, once again we have proved that D 0 for a rigid body motion.
0 ij
2 CONTINUUM KINEMATICS
215
&
Let us consider two line elements dX (1) and dX ( 2) in the reference configuration that
&
define the area element dA , (see Figure 2.26). After motion, theses vectors are transformed
&
&
&
into dx (1) and dx ( 2) , thus defining the new area element da , (see Figure 2.26).
&
da
&
J F T dA
&
dA
X 3 , x3
&
dA
R
&
dX ( 2)
&
dA
I 1 , e 1
&
da
Rc
&
da
&
dX (1)
&
dx ( 2 )
&
Qc
(1)
Pc dx
Current configuration - t
Reference
configuration - t
I 3 , e 3
&
da
I 2 , e 2
t1
X 2 , x2
X 1 , x1
The area element dA can be found using the definition of the vector product (the cross
product) used in Chapter 1, i.e.:
& &
dA( X )
PQ PR
&
&
&
dX (1) dX ( 2 )
&
dA N
dA N
(2.275)
&
where dA { dA is the magnitude of dA , and N is the unit vector which is normal to the
&
area element, i.e. codirectional with dA . In indicial notation, the cross product is
represented via the permutation symbol as:
dAi
. ijk dX (j1) dX k( 2)
(2.276)
o
o
&
&
P cQ c P cR c dx (1) dx ( 2)
&
&
&
da n da n
(2.277)
&
where da { da is the magnitude of da , and n is the unit vector associated with the da direction. In indicial notation, the deformed area element can be expressed as:
da i
. ijk dx (j1) dx k( 2)
(2.278)
216
& &
da ( x , t )
&
&
dx (1) dx ( 2 )
& (1)
&
F dX F dX ( 2 )
&
&
cof ( F ) (dX (1) dX ( 2 ) )
&
F F T dA
&
J F T dA
. ijk dx (j1) dx k( 2)
da i
(2.279)
In the above demonstration, by using tensor notation, we applied the tensor cofactor
&
&
definition, i.e. given a tensor T and two vector a , b , it holds that
&
&
&
&
1
T
cof ( T ) (a b) T a T b . Then, T 1 T >cof ( T )@ , (see Chapter 1). We also proved
in the same chapter that . rjk Frt F jp Fkq F . tpq , with which the equation in (2.279), in
indicial notation, can be expressed as:
. rjk Frt F jp Fkq Fti1 dX p(1) dX q( 2)
da i
J Fti1 dAt
(2.280)
&
J dA F 1
&
J F T dA Nansons formula
(2.281)
which is known as Nansons formula, and can also be written in terms of N and n as:
&
da
da n
J F T N dA
J dA N F 1
(2.282)
1 T &
F da
J
(2.283)
1 &
da F
J
J dA N F 1
(2.284)
Using equations (2.281) and (2.283), the magnitudes of da and dA are interrelated by:
da 2
& &
da da
dA 2
&
&
J 2 F T dA F T dA
&
&
J 2 dA F 1 F T dA
B C 1
J 2 dA 2 N B N
& &
dA dA
&
1 &
da F da F
2
J
&
1 &
da
F
FT da
J2
b
(2.285)
1
da 2 n b n
J2
Then:
da
J dA N B N
dA
1
da n b n
J
(2.286)
J 2 N B N
J2
1
(n b n )
Taking into account the equation in (2.282) the expression of n is obtained as:
(2.287)
2 CONTINUUM KINEMATICS
217
&
J dA F 1
&
J F T dA
J dA N F 1
J dA F T N
da
da
J dA N B N
J dA N B N
(2.288)
or
n
N F 1
F T N
N B N
N B N
(2.289)
N F 1
N B N 2
N F 1 l
N B N
1
2
N F 1
1
2
N B N
N F 1
N B N
N B N
3
2
N F D F
1
3
2
T
(2.290)
or
D (n )
Dt
n l n n D n
W
n
n l n n l n n
n l n n l n
>
n l n 1 l
2.9.1.1
(2.291)
@ n
&
dA N
& &
da ( x , t )
&
da n
&
JF T dA
(2.292)
&
The material time derivative of the deformed are element, da , is given by:
D &
(da )
Dt
>
&
D
J F T dA
Dt
&
DJ T
D
F dA J
F T
Dt
Dt
>
@ dA& J F
&
&
( x& v ) J F T dA l
&
&
JF T dA 1
&
da
T
> @
D &
dA
Dt
&
(2.293)
da
Then:
& &
&
D &
(da ) ( x& v ) da l T da
Dt
&
&
Tr (D) da l T da
&
Tr (D) 1 l T da
>
(2.294)
218
2.9.2
&
&
Let us consider a parallelepiped formed by the line elements dX (1) , dX ( 2) , dX (3) , (in the
reference configuration) whose volume is denoted by dV0 . After motion, the vectors
&
&
&
&
&
&
dX (1) , dX ( 2 ) , dX (3) are transformed into the line elements dx (1) , dx ( 2 ) , dx (3) , respectively,
and describe a new parallelepiped volume denoted by dV , (see Figure 2.27). Next, we can
&
&
establish the relationship between dV0 ( X ) and dV ( x , t ) .
dV
J dV0
dV0
dV0
dV
X 3 , x3
R
S
&
dA
I 3 , e 3
dX
dX ( 2)
(3)
I 1 , e 1
dx (3)
&
da
Rc
dX (1)
dx ( 2 )
Reference configuration - t
dV
Sc
I 2 , e 2
Pc
Qc
dx (1)
Current configuration - t
X 2 , x2
X 1 , x1
dX&
dV0
(1)
&
dX ( 2 )
dX&
( 3)
(2.295)
dx&
(1)
&
dx ( 2 )
dx&
( 3)
(2.296)
dx&
&
&
dx ( 2 ) dx ( 3 )
. ijk dxi(1) dx (j2) dx k(3)
(1)
dV
( 2)
( 3)
(1)
dV0
( 3)
T
(1)
(1)
( 2)
(2)
(2)
(2)
1
( 3)
( 3)
(2.297)
( 3)
Thus proving that the relationship between dV and dV0 is given by:
dV
F dV0
(2.298)
2 CONTINUUM KINEMATICS
219
&
&
S 0 dV0
dm S dV
(2.299)
dm
S 0 dV0
S dV
(2.300)
S0
S
dV
F J
dV0
&
&
S 0 ( X ) J S ( x, t )
(2.301)
Problem 2.17: Obtain an equation for mass density in terms of the third invariant of the
right Cauchy-Green deformation tensor, i.e. S 0 S 0 III C .
Solution:
From the equation in (2.301) we obtain:
&
&
S 0 ( X ) S ( x , t) J
and considering that the third invariant III C
J
III C , then:
S0
2.9.2.1
det (C ) det ( F T F )
J 2 , we obtain
(2.302)
III C
D
D
D
( JdV0 ) dV0
(J ) J
(dV0 )
Dt
Dt
Dt
(2.303)
Then we have:
D
(dV )
Dt
DJ
dV0
Dt
&
J x& v dV0
&
&
x v dV
Tr (D) dV
(2.304)
220
2.9.2.2
Dilatation
&
DV ( X , t )
If we considering that dV
2.9.2.3
(2.305)
F dV0
dV dV0
dV0
&
DV ( X , t)
dV dV0
dV0
J dV0 dV0
dV0
J 1
(2.306)
If during motion the volume element remains unchanged this implies that the Jacobian
determinant field is unitary, since:
F J1
dV
dV0
dV
D
( dV ) 0
Dt
dV0
(2.307)
Then, if during motion the volume of every particle remains unchanged the motion is said
to be isochoric, i.e.:
&
&
(2.308)
D
( JdV0 )
Dt
DJ
dV0
Dt
&
J x& v dV0
,
,
z0
&
x& v 0
(2.309)
z0
where we have taken into account that dV0 z 0 , J z 0 . By using the equation in (2.102), it
is possible to express the incompressibility of the form:
&
x& v v k ,k
&
Tr ( x& v )
Tr ( l )
Tr (D) 0
(2.310)
NOTE: It is interesting to point out that incompressibility is an approximation. That is, all
continuous media are compressible, but depending on the material, such as liquids in
general, the compressibility can be insignificant.
2 CONTINUUM KINEMATICS
&
)( X ) C
&
221
& &
G( x , t ) ) ( X ( x , t )) c
(2.311)
&
w) ( x , t ) wX k
wX k
wx i
&
wG( x , t ) 1
Fki
wX k
&
The normal to the surface ) ( X ) C is given by:
&
X& ) ( X )
N
&
X& ) ( X )
&
F T X& ) ( X )
&
x& G( x , t )
(2.312)
(2.313)
&
X& ) ( X )
(2.314)
&
x& G( x , t )
&
x& G( x , t )
(2.315)
Reference configuration
N
S0
Current configuration
x2
x3
x1
222
Material volume
t
Control volume
Control surface
0
v0
XP
X*
Material volume
Control volume
Control surface
t1
&
v( x * , t1 )
xP
x*
Material volume
Control volume
Control surface
t2
&
v( x * , t 2 )
xP
x*
&
&
&
&
D &
C G dx G Dt (dx) C G dx Gl dx C >G 1 Gl dx @
(2.316)
2 CONTINUUM KINEMATICS
223
&
&
&
D &
G da G Dt (da ) G da GTr(D)da l
&
D
G da
Dt S
da
&
(2.317)
&
>G 1 GTr(D)1 Gl da @
T
&
&
&
&
dV Dt G( x, t ) G( x, t ) Dt (dV )
dV Dt G( x, t ) G( x, t )
&
x
v dV
&
&
&
Dt G( x, t ) G( x, t )
&
x
(2.318)
v dV
&
Transport
Equations
a) Material curve
&
D
G dx
Dt C
G 1 Gl dx
C
b) Material surface
&
D
G da
Dt S
&
G 1 GTr(D)1 Gl da
&
&
D
G ( x , t )dV
Dt V
c) Material volume
(2.319)
&
&
Dt G( x, t ) G( x, t )
&
x
v dV
&
Next, we can apply the material time derivative to the equation in (2.319)(c) to obtain:
D
&
D
G( x , t )dV
Dt V
&
&
Dt G( x, t ) G( x, t )
&
x
v dV
&
&
&
&
&
wG( x , t ) & &
w
&
wt G ( x , t ) wx& v ( x , t ) G( x , t ) x v dV
V
&
&
&
&
wG( x , t ) &
w
&
wt G ( x , t ) dV wx& v G( x , t ) x v dV
V
V
V
(2.320)
&
wt G( x, t ) dV >
&
x
&
(Gv ) @dV
Now we apply the divergence theorem to the second integral on the right side of the
equation and we obtain:
flux of G acrossing
the surface S
&
D
G ( x , t )dV
Dt V
&
&
wG( x , t )
dV (,
Gv ) n dS
wt
V
S flux of G
(2.321)
local
As we can verify the volume integral on the right of the equation is a control volume and
&
the surface integral is a control surface. The term (Gv ) , (as discussed in Chapter 5),
represents the property flux ( G ) that crosses the control surface. We can also see that only
&
the normal component of the flux ( q n ) crosses the surface, whereas the tangential
224
component remains on the surface. Moreover, when the property is mass density, the
equation in (2.320) is known as the mass continuity equation, which is also discussed in
Chapter 5.
&
(Gv )
control volume
S
V
&
qn
&
wG( x , t )
wt
&
x
>(Gv& ) n @ n
control surface
& &
&
v ( x, t) dx
(
Circulation around (
(2.322)
The line integral (2.322) is denoted by the circulation around ( , (Chadwick (1976)) and by
using the Stokes Theorem we can obtain:
C (( )
& &
&
&
v ( x, t ) dx
(
&
( v ) n d8
8
&
8 n d8
&
x
(2.323)
&
where is the vorticity vector defined in (2.87) for which the physical interpretation in
(2.323) follows. Let us consider the region 8 , which consists of molecules subject to
circulatory motion, (see Figure 2.31). The equation in (2.323) ensures that the vorticity
contribution of all molecules in the region 8 is equal to the total circulation of the circuit
( (the boundary of 8 ). We can also verify that if the circulation around any closed curve
&
& &
is zero, it holds that x& v 0 . If this is the case, then the flow is said to be irrotational.
The rate of change of circulation around ( can be obtained directly from (2.319)(b) by the
&
& &
&
following change of variables G m ( x, t ) , da d8 , so, we obtain:
>
&
&
& &
&
D & &
( x , t ) d8
Tr (D) l T d8
Dt 8
8
& &
&
Tr (D) l
The motion is said to be circulation preserving if and only if
D
C (( )
Dt
(2.324)
T
&
0 , where
2 CONTINUUM KINEMATICS
& &
&
Tr (D) l
225
& &
& &
Tr ( l ) l 0
& & J
& &
F F 1 0
J
&
& J
& &
( JF 1 )
( JF 1 ) ( JF 1 ) F F 1 0
J
& &
1 &
1 &
JF JF J F 1 0
&
&
D
JF 1 0
Dt
(2.325)
&
JF 1
&
&
1
F 0 Cauchys vorticity formula
J
(2.326)
which is known as Cauchys vorticity formula, and links the vorticity between the reference
and current configurations with the circulation preserving motion, (Chadwick (1976)).
& &
v ( x, t )
&
d8
n
&
x3
x2
x1
226
~
F F vol
F iso F vol
(2.327)
~
where F { F iso is the transformation characterized by a volume-preserving (isochoric)
transformation, and F vol characterizes a volume-changing (dilatational) transformation,
1
3 F
F vol
(2.328)
J 31
&
X
pure dilatation
F vol
reference
configuration
B0
~
F
current
configuration
F
&
X
F
~
F F vol
1
~
3F
&
x
~
~
FT F
T
1
J 3 F
J
J
1
3
1
3
1
3 F
C vol
1
3
F
F
1
3
1
J J 1
2
3 C
F vol F vol
T
1 1
J 3 1 J 3 1
2
31
(2.329)
2
3 b
~
b ( J 3 1) b
~
b vol b
(2.330)
2 CONTINUUM KINEMATICS
227
where b vol
2
J
3
2J
2
3
5
3 JC
J
2
3 C
2
J
3
5
3 JC
J
5
3 J
2
3 C
(2.331)
2 F T D F , the equation in (2.331)
2
3
Tr (D) C
, J
2F T D F
F T F
T
T
Tr (D) F 1 F F D F 2 J
3
2
3 FT
~
2 F T D dev F
>
2J
0
2
3 F F 1D dev F F 1
ki iq
kp
pj jq
(2.332)
D 1 Tr(D)1 F
1
Tr (D)1 D dev . In addition, it holds that:
3
2 T
dev
1
T
2 J 3 F D F : F F
~
C : C 1
2
3 C
J Tr (D) and C
2
J
3
2J
2
3
2
1 1
dev
2 J 3 Fki D kp F pj Fiq F jq
E kq D dev
kp E pq
>
2J
2
3 D dev
qq
2J
2
3
@
(2.333)
Tr (D dev )
1
I ~ Tr F~ Tr J 3 F J 3 Tr F J 3 I
F
2
2
2 2
1 3 2
1 2
1 2
~2
3 F
3
~
~
I
I
I
Tr
F
J
I
Tr
J
J
I F Tr F 2
(
)
F
F
F
2
2
2
3
1
1
~
III F~ det ( F ) det J 3 F J 3 det ( F ) J 1 J 1
>
>
2
3
II F
(2.334)
and
1
1
I F vol Tr F vol Tr J 3 1 3J 3
2
2
2
2
1 2
1 3
9 J 3J 3 3 J 3
I F vol Tr ( F vol )
II F vol
2
2
vol
3
III F vol det ( F ) det J 1 J
(2.335)
228
2
Tr J 3 C
I C~
~
Tr C
II C~
~
1 2
I ~ Tr (C 2 )
2 C
III C~
~
det(C ) det J
>
2
3 I
1
J
2
IC
4
3 I2
C
(2.336)
III C
4
Tr J 3 C 2
4
3
II C
II C
III C2
(2.337)
2
3 C
2
J 3 det (C )
J 2 J 2
(2.338)
Likewise, we obtain:
I b~
Ib
3
III b
II b~
II b
3
III b2
I C~
III b~
I C , II b
;
(2.339)
II C , III b
II b~
II C~
III b~
III C~
(2.340)
2x2 x
yL
x 1 y C | y L
x 1
x
x
1
( J J T J T J ) . If the material displacement gradient components J ij are
2
2 CONTINUUM KINEMATICS
229
much smaller than the unity, i.e. J ij 1 , the quadratic terms ( J T J ) are even smaller and
can be discarded.
There are many cases in engineering in which the displacement gradient components are
small compared with the unity:
J ij
wu i
1
wX j
(2.341)
The approximation in (2.341) takes place when the continuum (structure) is made up of
very rigid material and the loads (forces) to which theses structures are subjected produce
small displacements. In such cases, motion can be approximated by the infinitesimal strain
theory, also known as the small deformation theory, or the small displacement theory.
1 wu i wu j wu k wu k
2 wX j wX i wX i wX j
small deformation L 1 wu i wu j
o E ij |
regime
2 wX j wX i
(2.342)
or in tensorial notation:
>
1 &
u X& X& u& 1 X& u& ( X& u& ) T
2
2
& &
& &
sym
sym u( X , t ) { X& u( X , t )
EL
>
1
J JT
2
(2.343)
If we now consider the Almansi strain tensor in terms of the gradient of the displacements,
(see Eq. (2.128)), we can define the linear Almansi strain tensor, eijL , as:
1 wu i wu j wu k wu k small deformation L 1 wu i wu j
regime
o eij |
2 wx j wxi
2 wx j wxi
wxi wx j
eij
(2.344)
or in tensorial notation:
eL
>
&
&
&
1 &
u x& x& u 1 x& u ( x& u) T
2
2
& &
& & sym
sym u( x , t ) { > x& u( x, t )@
@ 12 j j
T
(2.345)
NOTE: To verify that the material time derivative of the linear Almansi strain tensor e L is
equal to the rate-of-deformation tensor:
eijL
1 w wu i wu j
2 wt wx j wxi
1 w wu i w wu j
2 wt wx j wt wxi
1 w wu i
w wu j
2 wx j wt
wxi wt
1 w
w
vi
v j D ij
2 wx j
wxi
(2.346)
230
If both the displacement gradient and the displacement are small, it means there is very
little difference between the spatial and material configurations, so the linear strain tensors
can be considered equal, i.e.:
& &
& &
&
E L ( X | x, t ) | e L ( x | X , t ) ( x , t )
&
sym u
(2.347)
(2.348)
1
J JT
2
1 wu i wu j
2 wx j wxi
wu1
1 wu1 wu 2 1 wu1 wu 3
H ij H12 H 22 H 23
2 wx
2 wx3 wx 2
wx1
wx 2
H13 H 23 H 33 2
1 wu
wu 1 wu 2 wu 3
wu 3
1 3
wx3
>
@ >
&
&
1 &
1 &
u (u) T u (u) T
2
2
&
&
sym u skewu
(2.350)
where the symmetric part is the infinitesimal strain tensor, and the antisymmetric part is
known as the infinitesimal spin tensor, which is also called the infinitesimal rotation tensor. For
rigid body motion the condition that strain tensor is zero, i.e. 0 must be satisfied and
for motion characterized only by strain 0 must be true.
Notice that, the tensor does not accurately measure strain, since it is affected by rigid
body motion. To illustrate this, let us consider that a material body is subjected to a
rotation as indicated in Figure 2.34. In this situation, the equations of motion are given by:
x1
x2
x
3
cos T sin T 0 X 1
sin T cos T 0 X
2
0
0
1 X 3
x1
x2
x
3
X 1 cos T X 2 sin T
X 1 sin T X 2 cos T
(2.351)
X3
X 2 , x2
X2
x1*
x 2*
X1
X 1 , x1
2 CONTINUUM KINEMATICS
u 2
u
3
231
x1 X 1
X 1 (cos T 1) X 2 sin T
x2 X 2
X 1 sin T X 2 (cos T 1)
x3 X 3
(2.352)
If we consider the equation in (2.349) we can obtain the components of the infinitesimal
strain tensor as:
H ij
wu j
1 wu i
2 wX j wX i
0
0
(cos T 1)
0
(cos
1
)
0
T
0
0
0
(2.353)
As we can see H11 and H 22 are not equal to zero, but with small rotations it is true that
cos T | 1 , thus the terms H11 and H 22 are insignificant. As discussed in the chapter on linear
elasticity, the small deformation approximation is widely used for various engineering
problems and such problems are subjected to small displacements and small rotations. Note
that, for the rigid body motion described in Figure 2.34, the Green-Lagrange strain tensor
components are equal to zero. For instance, the component E11 works out as:
wu
wu
wu1 1 wu1
2 3
wX 1 2 wX 1
wX 1
wX 1
E11
(cos T 1)
>
1
cos T 12 sin 2 T
2
0 (2.354)
F m
M
1 2 M
O m 1
(2.355)
M 1
1 2 M
a x n
a n na n 1 x
1
1 x 2
1
n(n 1) n 2 2
a x
2!
(2.356)
1
1
x x2
2
8
In the event that x is very small, we can discard higher order terms, i.e.:
1
1 x 2
#1
1
x
2
(2.357)
Taking into account the Remainder above, the stretch and the unit extension, in a small
strain regime, are represented by:
O
m
F m
M
#1 M
M
1 2 M
O m 1
M
1# M
M
1 2 M
H (Nm )
(2.358)
which verifies that in a small strain regime the unit extension is equal to the normal
engineering strain.
232
1 2 N
M
1 2 M
M
N 1 2 N
N
2 M N
M
1 2 N N
1 2M M 1 2 N N
1 2M M
(2.359)
N
M
N
2M
N
M
N 2 M N
M
N
cos 4 2 M
(2.360)
where 4 is the angle between M and N , and is defined in the initial configuration. The
equation in (2.360) could have been obtained directly by applying the equation in (2.164)
and by considering that:
E N
cos 4 2 M
| cos 4 2 M N
O M O N
cos T
(2.361)
Additionally, if we consider that 'T T 4 is the angle variation, then the equation in
(2.360) can be rewritten as:
cos4 'T cos 4 2 M N
(2.362)
Moreover, the term on the left of the equation can be rewritten by the following
trigonometric relationship:
cos4 'T cos 4cos
'
T sin 4sin
'
T cos 4 'T sin 4
|1
| 'T
(2.363)
in which we have considered that the angle 'T is very small. If we substitute the previous
result into (2.362) we obtain:
N
cos 4 'T sin 4 cos 4 2 M
N
'T sin 4 2 M
(2.364)
Then the angle change for the small strain regime is given by:
'T
N
2M
sin 4
(2.365)
H11
H
12
H13
H 12
H 22
H 23
H13
H 23
H 33
H xx
H xy
H xz
H xy
H yy
H yz
H xz
H yz
H zz
(2.366)
2 CONTINUUM KINEMATICS
233
The stretch O x and the unit extension F x according to the x1 { x -direction can be
evaluated by considering the equations (2.358) and M e 1 , thus:
M
1 H
Ox #1 M
xx
Fx
(2.367)
O x 1 # H xx
Hence, we can conclude that the diagonal terms are related to the unit extension as follows:
H xx # F x
H yy # F y
e 1 and N
>1
H zz # F z
(2.368)
H xx
0 0@ H xy
H xz
H xz 0
H yz 1
H zz 0
H xy
H yy
H yz
H xy
X 2 , x2 { y
&
dX ( 2 )
M i
N
P
X 1 , x1 { x
N i
(2.369)
1
0
0
1
0
0
&
dX (1)
X 3 , x3 { z
2 M N
sin 4
1
'T xy
2
H xy
(2.370)
Afterwards we can interpret H xy as the angular distortion between two of the line elements,
whereas if we consider all three directions we obtain:
H xy
1
'T xy
2
H xz
1
'T xz
2
H yz
1
'T yz
2
(2.371)
"im
'x o0
u 'u u
'x
du
dx
(2.372)
234
'x
ti
Ac
Bc
t i 1
u 'u
Hx
wu
dx u
u
wx
dx
wu
wx
Hy
y, v
wv
v dy v
wy
dy
(2.373)
wv
wy
Bc
B
wv
v
dy
wy
dy
Ac
Oc
v
O
dx
u
wu
dx
wx
x, u
wu ( x, y, z )
wx
Hy
wv( x, y, z )
wy
Hz
ww( x, y, z )
wz
(2.374)
To find the tangential strain (or the shear strain), let us consider that the differential
element is only distorted by the angle shown in Figure 2.38. For small angles it holds that
tan T # T , then:
tan T1 # T1
wv
wx
tan T 2 # T 2
wu
wy
(2.375)
2 CONTINUUM KINEMATICS
J xy
T1 T 2
235
wv wu
wx wy
(2.376)
wv wu
wx wy
J yz
wv ww
wz wy
J xz
ww wu
wx wz
(2.377)
Note that 'T xy T 4 J xy , and if we compare that with the equation in (2.371) we can
conclude that J xy 2H xy . Similarly, we can say that J yz 2H yz and J xz 2H xz , thus,
H ij
H11
H
12
H13
H 12
H 22
H xx
H xy
H xz
H13
H 23
H 33
H 23
H xy
H yy
H yz
1
1
Hx
J
J
2 xy
2 xz
1
1
J
H
J
y
2 yz
2 xy
1 J xz 1 J yz
H z
2 2
H xz
H yz
H zz
(2.378)
Engineering Notation
Then, the strain components in terms of displacement in engineering notation are given by:
H ij
Hx
1
2 J xy
1 J xz
2
1
2
J xy
Hy
1
2
J yz
1
2
1
2
wu
wx
1 wu wv
2 wy wx
1 wu ww
2 wz wx
J xz
J yz
H z
wu
dy
wy
y, v
1 wu ww
2 wz wx
1 wv ww
2 wz wy
ww
wz
1 wu wv
2 wy wx
wv
wy
1 wv ww
2 wz wy
J xy
(2.379)
T1 T 2
Bc
B
T2
dy
T
Oc
Ac
T1
dx
wv
dx
wx
x, u
236
'V
dV dV0
O x dX 1 O x dX 2 O x dX 3 dX 1 dX 2 dX 3
1
(1 F x ) dX 1 (1 F x )dX 2 (1 F x )dX 3 dX 1 dX 2 dX 3
>(1 F x ) (1 F x ) (1 F x ) 1@dX dX dX
1
(2.380)
In a small strain regime, H ij 1 , the higher order terms can be discarded without there
being any significant change in the outcome. It is also true that with small strains unit
extensions are in keeping with the normal components of the strains F x1 H11 , F x2 H 22 ,
F x3
[F x F x F x ] dV0
1
[H11 H 22 H 33 ] dV0
(2.381)
&
DVL ( x , t ) { H V
H 11 H 22 H 33
Tr ( ) H1 H 2 H 3
(2.382)
H11 H 22 H 33
X3
(2.383)
dV
dV0
dxi
O x3 dX 3
dX 3
(1 F xi )dX i
O xi - stretch
dx3
X2
O x1 dX 1
O x2 dX 2
O xi dX i
F xi - unit extension
dX 1
dX 2
X1
dx1
dx 2
2.14.7
When the strain tensor field is independent of any one direction we say that the continuum
represents a plane strain state. In general, the independent direction is adopted by the x3 one. So, in this situation, the infinitesimal strain tensor components are given by:
H ij
H11
H
12
0
H12
H 22
0
0
0
0
Strain
Plane
o
H cij
H11
H
12
H12
H 22
(i, j 1,2)
(2.384)
The displacement field for a plane strain state is only a function of x1 and x 2 , i.e.:
u1
u1 ( x1 , x 2 )
u2
u 2 ( x1 , x 2 )
u3
C (constant)
(2.385)
2 CONTINUUM KINEMATICS
237
Problem 2.18: Consider a material body in a small deformation regime, which is subjected
to the following displacement field:
u1
(2 x1 7 x 2 ) u 10 3
(10 x 2 x1 ) u 10 3
; u2
; u3
x3 u 10 3
&
uij
wu1
wx1
wu 2
wx
1
wu 3
wx1
wu i
wx j
wu1
wx 3
wu 2
wx 3
wu 3
wx 3
wu 1
wx 2
wu 2
wx 2
wu 3
wx 2
7
0
2
1 10 0 u 10 3
0
0
1
m
m
>u& @
&
wu
wx&
m
.
m
Zij
skew &
1 wu i wu j
2 wx j wx i
u ij
0 4 0
4 0 0 u 10 3
0 0 0
0 1
0
1
b) The principal invariants are defined as I Tr ( ) , II
>Tr()@2 Tr ( 2 ) ,
2
III det ( ) , (see Chapter 1). Then, it follows that:
H ij
sym &
u ij
1 wu i wu j
2 wx j wxi
Tr ( ) (2 10 1) u 10 3
II
1
>Tr ()@2 Tr ( 2 )
2
III
det ( ) 11 u 10 9
11 u 10 3
2
3
0 2
3
0 2
3
0
6
3 10 0 3 10 0 3 10 0 u 10
0
0
1
0
0 1
0
0
1
1 u 10 6
0
0
1 H
H 3 11 u 10 3 H 2 H u 11 u 10 6 11 u 10 9
c) To draw the Mohrs circle for strain, (see Appendix A), we need to evaluate the
eigenvalues of . But, if we take a look at the components of we can verify that H 1 is
238
already an eigenvalue associated with the direction n i >0 0 r 1@ . So, to obtain the
remaining eigenvalues one only need solve the following system:
3
2 H
u 10 3
3
10 H
H
0 1
H 2
H 2 12 u 10 3 H 11 u 10 6
1.0 u 10 3
11.0 u 10 3
1.0 u 10 3
H II
1.0 u 10 3
11.0 u 10 3
H III
H I H III
2
6 u 10 3
H II
H III
1
2
J max
1
11
HI
H N (u10 3 )
H ij H ijsph
2
3 0 4 0
0
0 4 0 u 10 3
3
10
0
0
0
0
0
4
2 3 0
3 6 0 u 10 3
0 0 4
2 CONTINUUM KINEMATICS
239
F (1)
Intermediate configuration t1
F ( 2)
&
dx
Current configuration t 2
Initial configuration - t 0
X 3 , x3
&
dX
B0
&
dx
e 3
O
e 1
e 2
X 2 , x2
F ( 2) F (1)
X 1 , x1
&
&
&
&
&
F dX
;
dx F (1) dX
;
dx F ( 2 ) dx
(2.386)
&
By substituting the dx given by the second expression into the third one, we obtain:
&
&
&
&
dx F ( 2 ) dx F ( 2) F (1) dX F ( 2 ) F (1) dX
(2.387)
&
&
And by comparing (2.387) with dx F dX , we can conclude that:
&
dx
F ( 2 ) F (1)
(2.388)
1 T
F F 1 . We defined new tensors in a similar fashion because of the
2
transformations F (1) and F ( 2) , i.e.:
E
Initial configuration B0 :
&
E (1) ( X , t )
1 (1)T
F (1) 1
F
Intermediate configuration B :
&
E ( 2) ( x , t )
1 ( 2)T
F ( 2) 1
F
The Green-Lagrange strain tensor ( E ) can be written in terms of F (1) and F ( 2) as:
(2.389)
240
T
1 T
1 (2)
F F 1
F F (1) F ( 2 ) F (1) 1
2
2
T
T
1 (1)
F ( 2) F ( 2) F (1) 1
F
(2.390)
T
T
1 (1) T
F ( 2) F ( 2) F (1) 1 1 F (1)
F
2
2
T
T
1
F (1) E ( 2 ) F (1) F (1) F (1) 1
2 E ( 2)
2E ( 2) 1 ,
1 F (1) 1
(2.391)
E (1)
E (1) F (1)
E ( 2) F (1)
Thus, we can verify that E z E (1) E ( 2) , i.e. the Green-Lagrange strain tensor is not
additive for increments of motions. We can apply the same reasoning to the Almansi strain
tensor, e
1
1 F T F 1 , (current configuration). To demonstrate this, we define the
2
intermediate configuration B :
&
e (1) ( x , t )
T
1
1
1 F (1) F (1)
&
e ( 2) ( x, t )
T
1
1
1 F ( 2 ) F ( 2 )
(2.392)
The Almansi strain tensor can be written in terms of F (1) and F ( 2) as:
e
T
1
1
F ( 2) F (1)
1 F T F 1
1 F ( 2 ) F (1)
2
2
T
T 1
1
1
1 F (1) F ( 2) F ( 2) F (1)
1
(2.393)
T
T
1
1
1
1 F ( 2 ) F (1) F (1) F ( 2 )
T
1
1 2e , we obtain:
1 1 F
1 2e F
2
(1)
T
T
1
1
1
1 F ( 2 ) F (1) F (1) F ( 2 )
2
T
1
T
1
1
1 F ( 2 ) F ( 2 ) F ( 2 ) 2e (1) F ( 2 )
2
1
1
T
T
1
F ( 2 ) e (1) F ( 2 ) 1 F ( 2 ) F ( 2 )
F ( 2)
T
e (1) F ( 2)
1
( 2 ) T
(1)
( 2 ) 1
(2.394)
e ( 2)
Hence, we can see that e z e (1) e ( 2) and Figure 2.41 shows the strain tensors defined by
successive configurations.
2 CONTINUUM KINEMATICS
&
E (2) ( x , t )
&
e (1) ( x , t )
F (1)
241
F ( 2)
&
dx
&
E (1) ( X , t )
&
e (2) ( x , t )
&
dX
B0
&
dx
X 3 , x3
E (1) F (1)
e 3
O
E ( 2) F (1)
e
e 2
e 1
X 2 , x2
X 1 , x1
F ( 2)
T
e (1) F ( 2)
1
e ( 2)
F ( 2) F (1)
dL
L0
Lf
by integrating
o
H axial
L0
Lf
dL ln
L0
lnO
(2.395)
where H axial is known as the logarithmic strain or true strain. Note that, if there are successive
increments of displacement, i.e. L0 o L(f1) , and L(f1) o L f , it follows that the logarithmic
strain is additive, i.e.:
Lf
H Total
axial
1
dL
L
L0 0
L(f1)
f
L
L(f1)
1
1
ln f
dL
dL ln
L0
L(f1)
L
L
L0 0
L(f1) 0
(1)
( 2)
H
axial H axial
(2.396)
Then, from the logarithmic strain definition in (2.395), we can define the logarithmic strain
tensor as:
U ( Ln )
lnU
U ( Ln )
lnO N
a
(a )
(a )
N
a 1
(2.397)
Likewise, we can define this tensor in the current configuration, which is also known as the
Hencky strain tensor:
V ( Ln )
lnV
V ( Ln )
lnO n
a
(a)
n ( a )
a 1
The tensors V (Ln ) and U (Ln ) have the same eigenvalues. Also, the following condition is
satisfied:
242
Tr U ( Ln )
Tr V ( Ln )
lnO 1 lnO 2 lnO 3 lnO1 O 2 O 3
(2.399)
(a) N
(a)
1 N
a 1
(2.400)
1 O n
3
1
a
(a)
n ( a )
a 1
(2.401)
1 2
Um 1
C 1
m
m
E m (U)
1
lnU
lnC
2
for
mz0
(2.402)
for
m 0
where m is a positive integer number. The tensors defined above may be represented by
the eigenvalues of U (principal stretches), then:
1 3 m
(a) N
(a)
Oa 1 N
ma1
E m (O i )
3
(a) (a)
ln(O a ) N N
a 1
for
mz0
(2.403)
for
m 0
m 1
m 0
E ( 2 ) (U)
E (1) (U) H
E
(0)
1 2
U 1
2
1
C 1
2
(U) U
U 1
lnU
( Ln )
(2.404)
Next, we can define the strain tensors in the spatial configuration as:
1 m
V 1
e m (U) m
lnV
for
mz0
for
m 0
(2.405)
where m is a negative integer number. The above equations can be expressed in the
spectral representation as:
2 CONTINUUM KINEMATICS
1 3 m
O a 1 n ( a ) n ( a )
ma1
e m (O i )
3
(a)
(a)
ln(O a ) n n
a 1
for
243
mz0
(2.406)
for
m 0
2
1
m 0
e ( 1) ( V ) h
e
(0)
(V)
1
1 V 2
2
1
1 b 1
2
1 V 1
e ( 2 ) (U) e
lnV
( Ln )
(2.407)
'L
L0
L L0
L0
O 1
(2.408)
HH
L0
d"
"
L
ln
L0
ln(O)
(2.409)
1
(H C ) 2
2
(2.410)
or:
HH
ln(1 H C ) H C
&
&
dX 2 E dX
1 (ds ) 2 ( dS ) 2
2
(dS ) 2
HG
1D
(2.411)
L2 L20
2 L20
1 2
O 1
2
1
O 1O 1 1 O2 1
2
2
1
O 1 H C
2
HC
1 2
HC
2
(2.412)
&
&
dx 2e dx
1D
HA
1 (ds) 2 (dS ) 2
2
(ds ) 2
(2.413)
244
HA
L2 L20
2 L2
1
1 O 2
2
(2.414)
1 2
1
HC HC
2 1 H C 2
HG
1
O2
O2
HG
HA
(2.415)
'L
L
L L0
L
1 O1
(2.416)
HC
1 HC
(2.417)
L3 L30
3L20 L
1 2
O O1
3
(2.418)
(2.419)
We can draw a graph where the abscissa is represented by stretch ( O ) and the ordinate
represents the strains, (see Figure 2.42). We can also verify that for stretch values close to
unity the relations in (2.419) are met.
Engineering
Logarithmic
Green-Lagrange
Almansi
Swaiger
Kuhn
1
-1
-5
3 Stress
3
Stress
3.1 Introduction
When an external force is acting on a body, the atoms or molecules that make up the
continuum are affected and undergo a position change to achieve balance. Resistance to
this movement depends on the characteristics of the atoms or molecules that make up the
continuum. Internal resistance to movement is called internal force, and can be interpreted as
the average of the interatomic forces of a handful of atoms, thereby characterizing internal
force as a macroscopic variable. The internal force at each material point (particle) of the
continuum is represented by the traction vector field (force per unit area), which is the starting
point to establish the stress state at a material point.
3.2 Forces
When forces are applied directly to a body they are known as surface forces (e.g. contact forces
between two bodies), whereas when the body is immersed in, for instance, a gravitational
or electromagnetic field we have an indirect force. As regards forces, this chapter will only
deal with surface and gravitational forces.
3.2.1
An example of a surface force is illustrated in Figure 3.1 in which the water pressure on the
& &
dam is substituted by the surface force t * ( x ) also called the traction vector. The total force
acting on the dam wall can be obtained by means of the surface integral over the surface
S :
E.W.V. Chaves, Notes on Continuum Mechanics, Lecture Notes on Numerical
Methods in Engineering and Sciences 4, DOI 10.1007/978-94-007-5986-2_5,
International Center for Numerical Methods in Engineering (CIMNE), 2013
245
246
&
f
&* &
( x ) dS
&
df t
(3.1)
&
where df is the differential force acting on the differential area dS (surface element),
&
&
&
where it holds that df t * ( x ) dS . The unit of the surface forces in the International
System of Units (SI) is N / m 2 Pa (a Pascal is equivalent to one Newton per square
metre).
S
& &
t * ( x)
3.2.2
When the continuous medium is submerged in a gravitational field, the continuum is also
subjected to a force which can be represented either by force per unit mass (body force),
&
&
b , or by force per unit volume (force density), p . These two forces are related to each
other by means of the equation:
&
&
Sb i
Sb p
pi
(3.2)
& &
&
where S ( x , t ) is the mass density (mass per unit volume). The units of S , b , p , in the
&
kg
N
N
m &
, b
, >p@
. Then, the total
International System of Units (SI), are >S @
3
kg s 2
m
m3
>@
force acting on the body defined by the domain B can be evaluated by means of the
integral:
&
F
&
&
dF b dm
&
S b dV
(3.3)
3 STRESS
&
b i ( x, t )
247
0
0
g
m
s2
Hence, the total force acting on the body can be evaluated as follows:
Fi
&
Sb i ( x , t ) d V
S g dV
>m3 @
kg m m 3 s 2 dV
kg m
s2
N ( Newton ) .
x3
x2
x1
t
x3
x2
x1
B0
V0
S0
P0
Reference
configuration
St
Bt
&
P( x, t )
Vt
Current
configuration
248
3.3.1
3.3.1.1
Let us consider a continuum in the current configuration (deformed) which has been
&
divided into two parts by a plane 3 passing through the point P ( x , t ) , (see Figure 3.4).
This plane is defined by said point and by the normal n (unit vector). Let us also consider
&
a deformed area 'a centered at the point P ( x , t ) , so, the outcome of the internal force
o
acting on this area element is denoted by 'f . Then, we can define the traction vector (also
&
called the stress vector) at the point P ( x , t ) and which is associated with the normal n , as:
& &
t ( n ) ( x , t , n )
&
o
'f
"im
'a o0 'a
N
m2
Pa
(3.4)
Note that, the traction vector t (n) can vary from point to point and said variation defines the
&
traction vector field. Additionally, at a point P ( x , t ) the traction vector is only dependent on the
normal n . This traction vector represents the force per unit deformed area and its limit (3.4) exists
because the medium was assumed to be continuous.
Current configuration - t
Bt
'a o 0
&
t (n )
&
t (n )
&
x
x3
x2
x1
&
&
( 1)
point P ( x , t ) , thereby, defining the traction vector t ( n ) at the point, which is associated
with the normal n (1) , (see Figure 3.5(a)). Now if we consider a second plane 3 2 defined by
3 STRESS
249
&
the normal n ( 2) , which also passes through the point P ( x , t ) , there will be another traction
& ( 2)
vector t (n ) which is associated with this new plane, (see Figure 3.5(a)). Then an
immediate result of Cauchys Fundamental Postulate is the principle of action and reaction, (see
Figure 3.5(b)):
& &
& &
t ( x ,n ) t ( x , n )
(3.5)
& ( 2)
t (n )
n ( 2 )
& ( 1)
t ( n )
n (1)
32
31
31
& ( 1)
t (n )
& ( 1)
t ( n )
n (1)
P
n
a)
(1)
b)
Figure 3.5: Traction vector.
& &
&
The stress state at a point P ( x , t ) is completely described when the traction vector t ( x , n )
&
can be obtained for any arbitrary plane passing through this point P ( x , t ) . Cauchy showed
that if we define the traction vector on three mutually perpendicular planes passing through
&
the point P ( x , t ) we can fully describe the stress state at that point, (see Figure 3.6).
x3
e 3
&
t (e 3 )
&
t (e 2 )
&
t ( e1 )
x3
x1
Current configuration
x2
x2
e 2
e 1
x1
&
obtain three traction vectors associated with each direction and represented by t ( e1 ) , t (e 2 )
&
and t (e 3 ) respectively, (see Figure 3.6). Then by breaking each traction vector down
according to the directions x1 , x 2 and x3 , (see Figure 3.7), we can obtain:
250
&
t (e 1 )
&
(e )
t 2
&
(e 3 )
t
t 1e 2 e 1 t e2 2 e 2 t 3e 2 e 3
t 1e3 e 1
x3
t e2 3 e 2
(3.6)
t e3 3 e 3
&
t (e 3 )
t 3e3 e 3
&
t (e 2 )
t e2 3 e 2
t 1e3 e 1
t 3e 2 e 3
t 3e1 e 3
&
t ( e1 )
t e2 2 e 2
t e2 1 e 2
t 1e 2 e 1
x2
t 1e1 e 1
x1
&
&
&
t ( e1 ) e 1 t ( e 2 ) e 2 t ( e3 ) e 3
t 1e1 e 1 e 1 t 1e 2 e 1 e 2 t 1e 3 e 1 e 3
(3.7)
t e2 1 e 2 e 1 t e2 2 e 2 e 2 t e2 3 e 2 e 3
t 3e1 e 3 e 1 t 3e 2 e 3 e 2 t 3e3 e 3 e 3
Note that, in Chapter 1 we established that any second-order tensor can be represented by
a linear combination of dyads.
Then we can rearrange the components of into matrix form, and make a change in the
nomenclature so that we obtain:
t 1e 1
e 1
t 2
t e 1
3
t 1e3
t e2 3
t 3e3
t 1e 2
t e2 2
t 3e 2
V11
V
21
V 31
V12
V 22
V 32
V13
V 23
V 33
(3.8)
thus, defining V ij as the components of the Cauchy stress tensor also called the true stress tensor,
:
V ij (e i e j )
>Pa@
(3.9)
The representation of the Cauchy stress tensor components in the Cartesian system is
shown in Figure 3.8(a).
3 STRESS
251
x3
V 33
V 31
V 11
V 11
V 23
V 13
V 32
V 21
V 12 V 21
V 22
V 22
x2
V 12
V 31
V 32
V 23
a) front faces
x2
V 33
x1
x1
V 13
b) rear faces
V
21 V 22 V 23 V yx V yy V yz
V 31 V 32 V 33 V zx V zy V zz
Scientific
Notation
V x W xy W xz
W yx V y W yz
W zx W zy V z
(3.10)
Engineering
Notation
NOTE: It is important to note that many authors (mostly engineers) reverse the
convention of the indices, so to avoid misunderstandings, when we refer to engineering
notation the symmetry of the Cauchy stress tensor is already implicit.
z
Vz
W yz
W xz
W xz
Vx
W yz
W xy
Vy
W xy
252
Taking into account the symmetry of the Cauchy stress tensor, the representation of its
components in Voigt notation is given by:
V ij
3.3.2
V11
V
12
V
13
V13
Voigt
o^T`
V 23
V 33
V12
V 22
V 23
V11
V
22
V 33
V12
V 23
V13
V xx
V
yy
V zz
V xy
V yz
V xz
Vx
V
y
Vz
W xy
W yz
W xz
(3.11)
Our goal now is that given the nine components of the Cauchy stress tensor, how can we
find the traction vector associated with an arbitrary plane? It is very easy to answer this
question if we consider that the projection of the second-order tensor ( ) according to the
&
direction ( n ) is given by t ( n ) n , (see Chapter 1).
We will prove we can obtain the same result by starting from the forces equilibrium at the
material point. To do this, we define an arbitrary plane ABC with the normal n , (see
Figure 3.10), where the plane ABC passes through the point P .
*
dAT 12 ( AB AC )
*
(dAT ) i Sn e i
x3
Sn 1
*
dAT n { Sn
C
Sn 2
V 21
V 12
V 22
V 32
*
dA T
V 11
&
t (n )
V 31
V 13
V 23
A
V 33
x2
*
( dA T ) 1
Sn e 1
Sn1
*
(dAT ) 2
Sn e 2
Sn 2
AOC
*
( dA T ) 3
Sn e 3
Sn 3
AOB
ABC
BOC
Sn 3
x1
Associated with this plane is the traction vector t (n ) . Taking into account the Cauchy stress
tensor components in the rear faces of the tetrahedron, (see Figure 3.8(b)), and by
considering that the point is in equilibrium, the balance of forces according to the x1 direction is evaluated as follows:
(3.12)
3 STRESS
253
where S is the triangle ABC area and the projection of the area S according to the planes
x 2 x 3 , x1 x3 and x1 x 2 is given respectively by S n1 , S n 2 and S n 3 , (see Figure 3.10).
Then, if we simplify the equation (3.12) we obtain:
t 1( n)
(3.13)
Similarly, the balance of forces according to the directions x 2 and x3 provide us, the
following relationships respectively:
t (2n)
n1 V12 n 2 V 22 n 3 V 32
(3.14)
t 3( n)
n1 V13 n 2 V 23 n 3 V 33
(3.15)
Then by rearranging the equations (3.13), (3.14) and (3.15) into matrix form we obtain:
t 1( n )
( n )
t2
t ( n )
3
V11
V
21
V 31
V12
V 22
V 32
V13 n1
V 23 n 2
V 33 n 3
(3.16)
t i( n)
Tensorial notation
&
t ( n ) n
V ij n j
(3.17)
Then, by referring to the symmetry of the Cauchy stress tensor ( ), the traction vector
components can be represented in Voigt notation as follows:
t 1( n )
( n )
t 2
t ( n )
3
n1
0
0
n 2
0
0
0
n 3
n 2
n1
0
0
n 3
n 2
V11
V
n 3 22
V 33
0
V
n1 12
V 23
V13
^T `
>N @ ^T`
T
(3.18)
Problem 3.2: The Cauchy stress tensor components at a point P are given by:
V ij
x3
8 4 1
4 3 0.5 Pa
1 0.5 2
C (0,0,5)
n
&
x1
B (0,2,0)
x2
A(3,0,0)
254
3e 1 2e 2 0e 3
BA OA OB
o
e 1
e 2
0e 1 2e 2 5e 3
Then, the normal vector associated with the plane ABC is obtained by means of the cross
OC OB
BC
&
n
e 3
5 10e 1 15e 2 6e 3
0
&
Additionally, the unit vector codirectional with n is given by:
*
n 10
15
6
n
e 1 e 2 e 3
*
n 19
19
19
BC BA
0
3
2
2
Then by using the equation in (3.16), we can obtain the traction components as:
t (i n )
V ij n j
&
8 4 1 10
1
4 3 0.5 15 Pa
19
1 0.5 2 6
t1
t
2
t 3
t1
t 2
t 3
26
1
8 Pa
19
29.5
b) The traction vector t (n ) associated with the normal n can be broken down into a
&
&
normal ( N ) and a tangential ( S ) vector as shown in Figure 3.12. Then,
& (n ) &
&
&
t
N S
or
t ( n ) V N n V S s
&
&
where V N and V S are the magnitudes of N and S , respectively.
&
t (n )
x3
&
S
&
t (n)
&
t (n )
&
N
s
n
&
&
N S
&
N n
VN
&
t ( n)
V 2N V 2S
e 3
e 1
e 2
x2
x1
&
t ( n ) n ( n ) n n n : (n n ) t i( n) n i
(V ij n j )n i
n i V ij n j
V ij (n i n j )
Thus:
VN
t i n i
VN
10
1
>26 8 29.5@ 15 | 1.54 Pa
19 2
6
3 STRESS
&
t ( n )
V 2N V 2S
255
t i( n ) t i( n ) V 2N
V 2S
where
26
1
8 | 4.46
>
26
8
29
.
5
@
19 2
29.5
t i( n ) t i( n )
Thus,
VS
t i( n ) t i( n ) V 2N
Problem 3.3: The stress state at a point in the continuum is represented by the
components of the Cauchy stress tensor as:
V ij
2 1 0
1 2 0 Pa
0 0 2
a) Obtain the components of in a new system x1c , x c2 , x3c , where the transformation
matrix is given by:
x3
a ij
3 0 4
1
0 5 0
5
4 0 3
H1
xc3
xc1
where
a11
cos B 1
a12
cos C 1
a13
cos H 1
xc2
e 3
3
ec
e 1
2
ec
C1
1
ec
e 2
x2
B1
x1
256
3
1
0
52
4
These new components Vcij
Vcij
0 4 1 1 0 3 0 4
5 0 2 2 0 0 5 0
0 3 0 0 2 4 0 3
xc2
xc3
x2
P
x1
xc1
Tc A T A T
x3
V 33
xc3
V 23
V 13
V 13
2 0 .6 0
0 .6 2 0 .8
0 0.8 2
Vc33
Vc23
V 23
V 12
V 11
V 22
V 12
Vc23
Vc13
x2
Vc13
Vc22
xc2
Vc12
Vc12
Vc11
x1
xc1
T A T Tc A
Figure 3.13: Basis transformation.
b) The principal invariants of the Cauchy stress tensor can be calculated as follows:
I
II
III
Tr ( ) V ii
>
V11 V 22 V 33
1
1
V ii V jj V ij V ij
( Tr ) 2 Tr ( 2 )
2
2
2
2
V11V 22 V11V 33 V 33 V 22 V12
V13
V 223
det ( ) . ijk V i1V j 2 V k 3
1
V ii V jj V kk 3V ii V jk V jk 2V ij V jk V ki
6
2
2
V11V 223 V 22 V13
V 33 V12
3 STRESS
2 0
II
0 2
2 0
0 2
257
2 1
11
1 2
III
c) The principal stresses ( V i ) and principal directions ( n (i ) ) are obtained by solving the
following set of equations:
1
0 n1
2 V
1
2V
0 n 2
0
2 V n 3
0
0
0
0
V 3 I V 2 II V III 0
However, if we look at the format of the Cauchy stress tensor components, we can notice
that we already have one solution as in the x3 -direction the tangential components are
equal to zero, then:
V 3 2 Principal
direction
o n1(3) n (23) 0 , n 3(3) 1
To obtain the other two eigenvalues, one only need solve:
2V
2V
2 V 2 1
V 1 1
V 2 3
Then we can express the Cauchy stress tensor components in the principal space as:
1 0 0
V cijc 0 3 0 Pa
0 0 2
Additionally, the principal direction associated with V1 1 is calculated as follows:
1
0 n1(1)
2 1
1
2 1
0 n (21)
0
2 1 n (31)
0
with n 3(1)
0
(1)
(1)
0 n1 n 2
n (1) n (1)
2
1
0
2
n (21)
1
2
n1(1)
n (21)
1 we obtain:
then n i(1)
1
2
Since is a symmetric tensor, the principal space is formed by an orthogonal basis, so, it
is valid that:
n (1) n ( 2 )
n ( 3)
n ( 2 ) n ( 3)
n (1)
n ( 3) n (1)
n ( 2)
Thus, the second principal direction can be obtained by the cross product between n ( 3)
and n (1) , i.e.:
n ( 2 )
n ( 3) n (1)
e 1
0
1
e 2
0
1
e 3
1
0
1
1
e1
e2
2
2
1
23
0 n (22 )
0
2 3 n 3( 2 )
0
0
( 2)
( 2)
0
0 n1 n 2
(2)
n ( 2 ) n ( 2 ) 0 n1
2
1
0
n (22 )
258
With n 3(3)
n (22 )
then n i( 2 )
1 we obtain:
1
2
0
0
V 1 1
0
V2 3
0
0
V 3 2
0
1
2
1
2
1
2 1 0 2
1
0 1 2 0
2
0 0 2
1
0
2
1
2
0
V II
V III
(center )C1
Circle 2
(center )C 2
Circle 3
(center )C 3
1
(V II V III ) 1.5
2
1
(V I V III ) 2.0
2
1
(V I V II ) 2.5
2
(radius ) R1
(radius ) R 2
1
(V II V III ) 0.5
2
1
(V I V III ) 1.0
2
1
(V I V II ) 0.5
2
(radius ) R3
Then, we can illustrate the Cauchy stress tensor at P by means of Mohrs circle in stress as
shown in Figure 3.14.
VS
V S max
1
(V I V III ) 1.0
2
V S max
1
R2
R1
C3
V III
C1
V II
R3
VN
VI
3 V N max
3 STRESS
259
sph
Indicial notation
V ijsph V ijdev
V ij
dev
(3.19)
1
V kk E ij V ijdev
3
V m E ij V ijdev
V m 1 dev
V11 V 22 V 33
3
V1 V 2 V 3
3
1
V kk
3
I
3
1
Tr ( )
3
6
3
V m E ij
2 0 0
0 2 0
0 0 2
2E ij
1
V12
(2V 22
3
V
13
0
0
V m
V12
V 23
V11 V 22 )
V13
V11 V 33 )
V 23
1
3
(2V 33
Thus,
1
0 0 1
2 2
1
2
2
0 1 0
0
0
2 2 0 0
Now let us remember from Chapter 1 that and dev are
V ijdev
0
0
0
V1
0
0
dev
0
V2
0
V m
0
V 3 0
0
0
0
Vm
0
1 0 0
0 1 0
0 0 0
0
0
V m
Tr ( dev ) 0
1
II dev
III dev
Traditionally, in engineering, the invariants of the deviatoric stress tensor are represented
by:
J1
I dev
J2
II
J3
III dev
dev
1 2
I 3 II
3
1
2 I 3 9 I II 27 III
27
260
x3
V 33
V 23
V 13
V 13
V 23
V 12
V 11
V 22
V 12
x2
1
x3
x3
Vm
dev
V 33
V 23
V 13
V 13
Vm
x2
Vm
V 23
V 12
V dev
22
V 12
dev
V 11
x1
x2
x1
sph
dev
V oct
S { W oct
1
V1 V 2 V 3 1 V ii
3
3
1
2 I 2 6 II
3
2
J2
3
I
3
Vm
V V V
dev 2
1
dev 2
2
dev 2
3
3.3.3
3.3.3.1
Vm
W oct
2
J2
3
2
3
As we have seen before, the Cauchy stress tensor was derived in the current configuration
(deformed). In some cases we may wish to adopt the Lagrangian description for studying
3 STRESS
261
motion, and then it will be necessary to correlate the Cauchy stress tensor with a
hypothetical stress tensor in the reference configuration, (see Figure 3.16).
Reference configuration - t 0 (t
&
dA
P0
B0
&
da
Bt
& ( N )
t0
&
dA
X 3 , x3
Current configuration - t
0)
&
t (n )
&
da
X 2 , x2
X 1 , x1
In the reference configuration, we adopt an area element dA with the normal N and
&
associated with that plane we can define a pseudo traction vector t 0 ( N ) . After motion, this area
&
element becomes da in the deformed configuration, associated with which we have the
& (n )
traction vector t , (see Figure 3.16). Then by using the definition in (3.4) we can define
& ( N )
&
t0
and t (n ) , respectively, as:
&
&
& ( N ) &
& ( n ) &
'f df
'F d?
t 0 ( X , t , N ) "im
;
t ( x , t , n ) "im
&
&
(3.20)
'Ao0 'A
'a o0 'a
da
dA
&
df
d? i
df i
(3.21)
t (0N ) i dA t (i n) da
&
&
Pik N k dA V ik n k da
V ik da k
Pik dAk
(3.22)
(3.23)
&
&
J F T dA
J F dA
T
&
(3.24)
J F T
1
PFT
J
Pij
J V ik F jk1
V ij
1
Pik F jk
J
(3.25)
262
where P is the first Piola-Kirchhoff stress tensor which is also called the nominal stress tensor. This
tensor represents the force in the current configuration per unit undeformed area, so, it is
both a two-point second-order tensor and a non-symmetric tensor, i.e. P z P T .
&
&
&
F 1 t 0
&
t 40
&
RT t0
&
t*
&
Jt
(3.26)
where R is the polar decomposition rotation tensor, (see Chapter 2). In addition, the
following relationships are valid:
&
t0
&
t 0/
&
t 04
3.3.3.2
da &
t
dA
&
F 1 t 0
&
RT t0
&
F t 0/
&
R t 04
&
F 1 R t 04
&
R T F t 0/
da 1 &
F t
dA
da T &
R t
dA
(3.27)
We define the Kirchhoff stress tensor , which is related to the traction vector t * (current
configuration), as:
&
t*
&
Jt
J n
n
o J
(3.28)
As we can verify the Kirchhoff stress tensor is a symmetric second-order tensor and is
related to the Cauchy stress tensor and to the first Piola-Kirchhoff stress tensor by means
of the following relationships:
J ; P F T ; P F 1
3.3.3.3
(3.29)
We can also introduce the second Piola-Kirchhoff stress tensor, S , which is defined in the
reference configuration, as:
Tensorial notation
F 1 P
Indicial notation
F 1 F T
S ij
JF 1 F T
Fik1 Pkj
Fik1 W kl F jl1
J Fik1 V kl F jl1
(3.30)
or
P
F S
PFT
F S F T
(3.31)
The Cauchy stress tensor can be expressed in terms of the second Piola-Kirchhoff stress
tensor as:
1
F S F T
J
JF
1
F T
JF 1 F T
(3.32)
S:
(3.33)
3 STRESS
Reference configuration - t 0 (t
263
Current configuration - t
0)
Bt
B0
&
t (n ) da
& ( N )
t 0 dA
&
dA
&
t 0 P N
P -First Piola-Kirchhoff stress tensor
n
&
da
&
t n
- Cauchy stress tensor
&
t 0/ dA
B0
Bt
F 1
&
dA
&
t (n ) da
& ( N )
t 0 dA
&
t 0/ S N
S -Second Piola-Kirchhoff stress tensor
&
da
&
t*
&
Jt
n
- Kirchhoff stress tensor
&
t 40 dA
B0
RT
&
dA
& ( N )
t 0 dA
&
t 40 T N
T - Biot stress tensor
264
3.3.3.4
We can also introduce the Biot stress tensor, T , which in general is non-symmetric and is
&
related to the traction vector t 40 :
&
t 40
T N
(3.34)
&
RT t 0
(3.35)
R T P N
Thus,
T RT P
(3.36)
J R T F T
(3.37)
If we refer to the equations in (3.36) and (3.31), the tensor T can also be expressed as:
T RT P RT F S RT R U S U S
(3.38)
where we have applied the right polar decomposition F R U . Note that the tensor T
will be symmetrical if U and S are coaxial tensors, hence S U U S (U S) skew 0 ,
(see Chapter 2 in subsection 1.5.9 Coaxial Tensors).
3.3.3.5
We can also introduce the Mandel stress tensor denoted by M , which in general is nonsymmetric and is defined as:
M C S
F F 1 P
F T P
F T F T
(3.39)
where C is the right Cauchy-Green deformation tensor, which was defined in Chapter 2.
We can now summarize the relationships between stress tensors defined above as:
1
PFT
J
J F T
F 1
J PFT
S
J F 1 F T
J R T F T
M J F T F T
1
F S F T
J
1
R T FT
J
F S R T
F S F T
F 1 P
F T M
R T FT
F 1 F T
R T P R T F T
F T P
F T F T
1 T
The Cauchy stress
F M F T
(3.40)
tensor
J
F T M F T
U 1 T C 1 M
U S U 1 M
C S U T
(3.41)
(3.42)
(3.44)
(3.45)
3 STRESS
F | F 1 | 1
and
265
det ( F ) | 1
(3.46)
Reference
configuration
B0
Current
configuration
J F T
&
X
Ref. Conf.
Current Conf.
JF 1 F T
&
S( X , t )
&
T( X , t)
&
M( X , t ) C S
&
x
M JF
JR
&
( x, t )
&
( x, t ) J
F
F T
T
3.3.4
For the next stress tensor representation let us consider that the Cauchy stress tensor ( )
and the left stretch tensor ( V ) are coaxial, i.e. they present the same principal directions.
Then, the spectral representation of the Cauchy stress tensor is given by:
n ( a ) n ( a )
(3.47)
a 1
(a)
n ( a ) N
F 1
a 1
1 (a ) (a )
N n
O
a 1
F T
1 (a) (a)
n N
O
a 1
(3.48)
and also that the polar decomposition rotation tensor is represented by:
R
(a)
(a )
N
(3.49)
a 1
J J V a n ( a ) n ( a )
a 1
a 1
JV a n ( a ) n ( a )
W
a 1
n ( a ) n ( a )
(3.50)
266
3
3 1 (a ) (a )
J V a n ( a ) n ( a )
n N
a 1
a 1 Oa
J F T
3
a 1
JV a ( a ) ( a ) ( a ) ( a )
n n n N
Oa
JV a ( a ) ( a )
n N
1 Oa
(3.51)
(a)
Pa n ( a ) N
a 1
As we can verify the first Piola-Kirchhoff stress tensor is neither in the current nor in the
reference configuration, i.e. P is a two-point tensor, and Pa are not the eigenvalues of P .
J F 1 F T
3 1 (a)
3
3 1 (a) (a)
N n ( a ) V a n ( a ) n ( a )
n N
J
a 1 Oa
a 1
a 1 Oa
3
JV a ( a )
a
a
a
a
a
(
)
(
)
(
)
(
)
(
)
N n n n n N
2
a 1 Oa
JV a ( a ) ( a )
N N
2
O
a 1
(3.52)
(a ) N
(a)
N
a 1
As we can verify, the second Piola-Kirchhoff stress tensor has the same principal directions
as the right stretch tensor ( U ).
J R T F T
J
a
(a)
US
3
3 1 (a) (a )
n ( a ) V a n ( a ) n ( a )
n N
a 1
a 1 Oa
JV a ( a ) ( a )
N N
O
a 1
3
(a) N
(a)
N
a 1
3
3
(a ) N
( a ) S N
(a) N
( a )
Oa N
a
a 1
a 1
aO a
(3.53)
(a) N
(a )
N
a 1
J O a N
a
O n N
a 1 a
a 1
a 1
JV
(a) N
(a )
N
(3.54)
(a ) N
(a)
N
a 1
a 1
J
Va
O2a
1
Wa
O2a
1
Pa
Oa
1
Ta
Oa
1
Ma
O2a
(3.55)
3 STRESS
267
J dev F T JV m F T
JF 1 dev F T JV m C 1
where P and S are the first and second Piola-Kirchhoff stress tensors, respectively, C is
the right Cauchy-Green deformation tensor, F is the deformation gradient, J is the
Jacobian determinant, and the scalar V m is the mean normal Cauchy tress. Also prove that
the following relationships are true:
P:F
S :C
3JV m
Solution:
First of all we prove that P : F S : C :
P:F
Pij Fij
( Fik S kj ) Fij S kj ( Fik Fij )
S kj ( F T F ) kj S kj (C ) kj
S :C
J dev F T JV m 1 F T
J dev F T JV m F T
Thirdly, by taking into account the definition S JF 1 F T , (see equation (3.30)), and
by breaking down into sph dev , we obtain:
S
JF 1 F T
S ij
JFik1 V kp F jp1
JF 1 ( dev V m 1) F T
1
JFik1 (V dev
kp V ( m ) E kp ) F jp
JF 1 dev F T JF 1 V m 1 F T
1
1
1
JFik1 V dev
kp F jp JFik V ( m ) E kp F jp
JF 1 dev F T JV m ( F T F ) 1
1
1 1
JFik1 V dev
kp F jp JV ( m ) Fik F jk
JF 1 dev F T JV m C 1
1
1
JFik1 V dev
kp F jp JV ( m ) C ij
Then by applying the double scalar product between S and C we can obtain:
S :C
( JF 1 dev F T JV m C 1 ) : C
JF 1 dev F T : C JV m C 1 : C
( JF 1 dev F T ) : ,
C
( JF 1 dev F T ) ij ( F T F ) ij
1
( Fip1V dev
pk F jk )( Fqi Fqj )
F T F
J E qp E qk V dev
pk
J V dev
pk E pk
J V dev
kk
dev
J
:1
Tr ( dev ) 0
Thus:
JV m Tr (C 1 C ) JV m Tr (1) 3 JV m
Now, by taking the double scalar product between P and F we obtain:
P : F J dev F T : F JV m F T : F
S :C
JV m C 1 : C
268
( J dev F T ) ij ( F ) ij
J V ikdev F jk1 Fij
dev
J
:1
JV ikdev E ik
0
Tr ( dev ) 0
Thus,
P:F
JV m F T : F
JV m Tr ( F T F T )
JV m Tr (1) 3 JV m
4. Objectivity of Tensors
4
The Objectivity of Tensors
4.1 Introduction
Any physical quantity must be invariant for different observers. For example, let us
suppose that two observers are located at different positions, (see Figure 4.1), this means
they must both detect the same stress state acting on the body for there to be physical
meaning.
observer 1
Current configuration
observer 2
observer
&
x*
&
&
c (t ) Q(t ) x
B*
Current configuration
269
270
The equivalent to the two observers is that one single observer that records the stress state
in the current configuration must be able to compute the same stress state if the continuum
undergoes rigid body motion.
When we are dealing with nonlinear problems it is necessary to approach the constitutive
equations in rates. As we shall see, in general, the rate of change of the tensor, e.g. velocity,
acceleration, etc., is not objective, which can be inconvenient when formulating the
constitutive equation, which by definition must be objective. Therefore, to overcome this
drawback, we will define some rates that are objective.
&
x*
F*
&
X
&
g*
G*
QF
A*
&
G
&
g
&
x
Reference configuration
F
G
A
Current configuration
(4.1)
271
Vectors
&
&
If g is an Eulerian vector which is generated by the motion F , then we can state that g is
&
&
objective if its counterpart g * , which is generated by F * , is related to g by means of the
equation:
&
g*
&
Q g
(4.2)
NOTE: We will take this opportunity to mention that the orthogonal transformation law
for two-point tensors (pseudo-tensors) is the same as that for vectors. As examples of twopoint tensor we can quote: the deformation gradient, F * Q F ; the first Piola-Kirchhoff
stress tensor, P * Q P ; and the Polar Decomposition rotation tensor, R * Q R .
&
As an example of an objective vector we can quote the area element vector da . To prove
that, let us consider Figure 4.3, where the area element in the rotated current configuration
is defined as:
&
&
F * dX (1) & F * dX ( 2 )
&
&
&
(1)
(Q F ) dX (Q F ) dX ( 2) Q dx (1) Q dx ( 2)
&
&
&
&
cof (Q)(dx (1) dx ( 2) ) Q (Q T )(dx (1) dx ( 2 ) )
&
Q da
&
Hence, we have demonstrated that da is objective.
&
da *
& (1)
& (2)
dx * dx *
(4.3)
& (2)
dx *
& (1)
dx *
F*
&
dA
&
da
&
dX ( 2 )
&
dx ( 2 )
&
dX (1)
&
dx (1)
Reference configuration
Current configuration
F
&
&
cFX
(4.4)
&
&
c F X
(4.5)
272
&
Qv
&
&
Q c Q F X
(4.6)
&
&
c F* X
&
x *
&
x *
&
x *
&
& D
c
(Q F ) X
Dt
&
&
c F * X
&
&
& D
D
c
( Q) F X Q
(F ) X
Dt
Dt
&
&
&
c Q
F X Q F X
(4.7)
If we compare the last line of the equation (4.7) with (4.6) we can conclude that the
& &
0.
velocity is not objective as the only way to achieve this is when c 0 and Q
Then the rate of change of (4.5) provides us the acceleration, i.e.:
x& { a&
(4.8)
(4.9)
&
&
&
&
& Q
F X Q
F X Q
F X Q F X
c
(4.10)
If we compare the equations (4.9) with (4.10) we conclude that the acceleration is not
& &
0.
objective as this will only be so if and only if c 0 and Q
Second-order tensors
&
(4.11)
Another special case is the two-point tensor which is neither in the current nor in the
reference configuration. In this case the orthogonal transformation is characterized by:
A*
QA
(4.12)
&
The Lagrangian second-order tensor, A ( X , t ) , is objective when the following condition is
satisfied:
&
A * ( X , t)
&
A( X , t)
(4.13)
Note that, the reference configuration has not been rotated, (see Figure 4.4).
4.2.1
As we saw in Chapter 2, the deformation gradient (two-point tensor) relates line elements
&
&
between reference and current configurations, i.e. dx F dX , (see Figure 4.4). With
components this relation becomes:
dxi
Fij dX j
wx i
dX j
wX j
(4.14)
273
Additionally, we can define the deformation gradient in the rotated current configuration
as:
Fij*
wx i* wx k
wx k wX j
wx i*
wX j
F*
Q ik Fkj
QF
(4.15)
&
&
dx *
&
Q F dX
&
Q dx
(4.16)
It is now simple to show, (see Figure 4.4), that the following relationship is valid:
F
&
dx *
&
F * dX
QT F *
QT Q F
&
dx *
&
Q F dX
F*
(4.17)
&
dx *
&
x*
QF
&
Q dx
QT
Reference
configuration
&
&
dX { dX *
B0
& &
X { X*
&
dx
&
dx
&
F dX
&
x
Current
configuration
F 1 Q T
(4.18)
F*
1
Q F 1
F*
J*
(4.19)
4.2.2
Kinematic Tensors
Taking into account the rotated current configuration, the right Cauchy-Green deformation
&
tensor ( C ( X , t) F T F ) is defined as:
C*
F*
F * Q F T Q F
F T QT Q F
FT F
(4.20)
274
Then it follows from the above equation that the Green-Lagrange strain tensor,
&
E( X , t)
1
C 1 , is given by the following transformation law:
2
E*
1 *
C 1*
2
1
C 1 E
2
(4.21)
Note that the tensors C and E are objective, since they are defined in the reference
configuration.
&
b*
Q F Q F T
Q F F T QT
Q b QT
(4.22)
We can find the same result if we start from the following definition, b F C F 1 , (see
Chapter 2), i.e.:
b*
F * C* F*
F* F
*T
1
b*
F* F
F * C F *
* 1
F* F
1
(Q b Q T ) 1
(Q T ) 1 b 1 Q 1
*T
Q b 1 Q T
1
Q F F T F Q F 1
Q F F T F F 1 Q 1
Q F F T QT
Q b QT
(4.23)
1
1 b 1
2
&
configuration becomes:
Q e QT
e*
(4.24)
Starting from the polar decomposition defined in Chapter 2, (see Figure 4.5), it follows
that:
R U V R
R * U*
F*
V* R*
(4.25)
&
where R is the polar decomposition (rotation tensor) proper orthogonal tensor, U( X , t ) is
&
the right stretch tensor, and V ( x , t ) is the left stretch tensor. Then by taking into account
&
that C ( X , t ) F T F U 2 and the equation in (4.20) we deduce that:
C*
U*
U2
U*
(4.26)
U
Q R U U
o
*
R*
Q R
Q R* RT
(4.27)
V*
Q b QT
b*
1
V*
2
Q b 1 Q T
(4.28)
V* R*
QF
V* R*
Q V R
V* R*
(4.29)
275
Moreover, if we bear in mind that Q R * R T , (see Eq. (4.27)), we can conclude that:
Q V R
V
V* R* Q V R RT
Q V Q
V* R* RT Q V
V* Q
(4.30)
R*
Q R
B * {B
F * C F *
b*
U*
1
B*
b*
Q b QT
e*
Q e QT
&
x*
QT R *
Rotated current
configuration
Q
F*
QF
QT
B0
Reference
configuration
&
x
&
X
C
U2
E*
C*
U*
Current
configuration
b, e , V
F C F 1
4.2.3
Stress Tensors
&
Q QT
(4.31)
& (n * )
t*
&
n t (n)
&
* Q n Q t (n)
&
Q T * Q n t (n)
& (n * )
t*
defined
(4.32)
276
The first Piola-Kirchhoff stress tensor, defined in Chapter 3, is given by the equation
P det ( F ) F T . Then, P can be defined in the rotated current configuration as:
det ( F * ) * F *
P*
Q Q and F
T
(4.33)
det (Q F )Q Q (Q F )
T
1
T T
det (Q)det ( F )Q Q ( F Q )
det ( F )Q Q T Q F T
Q det ( F ) F T
P
det ( F * ) * F *
P*
T
T
P*
Q P
(4.34)
Note that the first Piola-Kirchhoff stress tensor is a two-point tensor whose
transformation is defined according to the transformation law of vectors, (see Figure 4.6),
hence, the first Piola-Kirchhoff stress tensor is objective.
Rotated current configuration
F
QF
& (n * )
t * da
Bt*
Q P
P*
Q QT
n *
* J * *
B0
&
dA
& (N )
t 0 dA
Bt
Reference configuration
F
S S*
M M*
T
&
da
J F T
&
t (n) da
Current configuration
T*
J
Figure 4.6: Objectivity of the stress tensors.
&
* F *
1
T
T
T
det (Q F ) F Q Q Q Q F
det ( F ) F 1 F T
det ( F * ) F *
1
T
S*
(4.35)
277
Now, if we consider the Kirchhoff stress tensor, ( x , t ) J , it follows that in the rotated
current configuration we obtain:
* J * * J Q QT
Q J QT
Q QT
(4.36)
F*
P * Q F T Q P
T*
U* S *
US
&
T( X , t)
F T P
&
M( X , t )
(4.37)
F *
(4.38)
1
F * F *
Q F Q F F
l
F 1 Q T becomes:
T F 1 Q T
F 1 Q
* 1
1
(4.39)
QT
F F 1 Q T Q F F 1 Q T
Q
QT Q l QT
Q
(4.40)
(4.41)
Q T ) that appears in
Hence, the tensor l is not objective due to the additional term ( Q
as:
the equation (4.41) by means of which we can obtain Q
Q
Q Ql
D
Q QT
Dt
T
Q
QT l
*T
l
QT
(4.42)
1 , we obtain:
D
1 0 Q Q T
Dt
T
Q Q
(4.43)
T Q l QT
Q Q
Then if we use (4.44) we can obtain another way to express the rate of Q :
(4.44)
278
Q
Ql
l
*T
T
Q
Then by substituting
obtain:
* sym
1
l
2
{ D*
QT
QT l
(4.45)
l
*T
(4.46)
>
1
T Ql
Q QT Q l QT Q Q
2
D*
Q T @
(4.47)
>
1
Q l QT Q l
2
D*
QT @
>
1
l l
2
@ Q
Q D QT
D*
(4.48)
(4.49)
QT Q l QT
Q
Q T Q (D W ) Q T
Q
QT Q D QT Q W QT
Q
D* W *
(4.50)
W*
(4.51)
Q W QT W *
Q
QT
Moreover, if we also consider that Q
becomes:
QT
Q
4.3.1
T
QQ
Q W QT W *
W* Q Q W
Q Q
T T
T
Q
(4.52)
W QT QT W *
(4.53)
Objective Rates
&
&
Qa
(4.54)
&
&
a
Q
Q a
(4.55)
279
&
The above proves that the rate of change of a is not objective, since an additional term
&
a
(Q
) appears in the above equation.
As we have seen before, a second-order tensor that is defined in the current configuration
is objective if it holds that:
A*
Q A QT
(4.56)
*
A
(4.57)
Thus, we can conclude that A is not objective, since A * z Q A Q T . We can then define
some objective rates.
4.3.1.1
&
&
&
a
a * Q
Q a l
&
&
&
&
T
a * l * Q a Q l T a Q a
&
&
&
T &
a * l * a* Q a l T a
C
&
a*
*T
&
&
&
Q a Q l T a Q a
(4.58)
C
&
Qa
C
&
C
&
The rate ( x ) indicates the convective rate with which we can introduce a new vector rate a ,
which is objective and is defined as:
C
& &
a a l
4.3.1.2
&
a The convective rate
(4.59)
l * Q Q l and Q
T
If we use the equation given in (4.42), i.e. Q
and by substituting them into the equation in (4.57) we obtain:
*
A
Q Q l A Q T
Q A Q T Q A Q T l
*T
l
*T
l
Q T
QT ,
(4.60)
or
Q A QT Q A QT l *
* l * Q A Q T Q A Q T l *T
A
* l
A
A*
A * l
A*
A* l
*T
A*
Q l A Q T Q A Q T Q A l
l A Al
Q A
l A Al
Q A
QT
(4.61)
Q A Q T
A l A A l
(4.62)
280
Problem 4.1: Obtain the Oldroyd rate of the left Cauchy-Green deformation tensor ( b ).
Solution:
Based on the definition of the Oldroyd rate in (4.62), we can obtain the Oldroyd rate of b
as b b l b b l
l F F
becomes b F F T F F T
conclude that b b l b b l
4.3.1.3
F l F
b bl T
b b l T . Thus, we can
l b bl
0.
and Q
T given in (4.42), we use the equations
If instead of using the equation related to Q
given in (4.45), and by substituting them into (4.57) we obtain:
*
A
Q l
* l
A
*T
* l
A
*T
l
*T
Q A Q T
Q A QT Q A QT l *
Q
A QT Q A QT l *
A*
* l
A
'
A*
*T
QT Q A l QT QT l
QA
A*
A* A* l *
l
Q A
Ql
A Q T Q A Q T Q A l Q T
Q A l T A A l Q T
T
(4.63)
A A l Q T
'
Q A Q T
l
A
A Al
(4.64)
Problem 4.2: Obtain the Cotter-Rivlin rate of the Almansi strain tensor ( e ) in terms of the
rate-of-deformation tensor ( D ).
Solution: Based on the definition of the Cotter-Rivlin rate in (4.64), the Cotter-Rivlin rate of
'
e l
D e l
e e l
e D l
e el
'
'
W A A W QT W* A * A * W*
Q A
(4.65)
Q A W A A W Q T
$
A*
Q A Q T
(4.66)
We can conclude by the above that the rate A , called the Jaumann-Zaremba rate, is objective
and is given by:
A W A A W
281
(4.67)
'
Next, we can interrelate the rates A , A and A . To do this let us consider the following
equations:
l A Al
A
A
$
A
'
A
A l A A l
$
(4.68)
(4.69)
WA AW
A
A
A W A A W
A Al
A
A l
l
A
'
A Al
(4.70)
A W A A W
A l A A l
A l A A l
A D W A A D W W A A W
WA AW
(4.71)
Then, we can connect the Jaumann-Zaremba rate to the Oldroyd rate by:
$
A D A A D
(4.72)
A W A A W
A l
A Al
'
'
A l
A Al
WA AW
A D W A A D W W A A W
(4.73)
Then, we obtain the relationship between the Jaumann-Zaremba rate and the Cotter-Rivlin
rate:
'
A D A A D
(4.74)
Now by adding equations (4.72) and (4.74) we can reach the following conclusion:
$
2A
'
(4.75)
A A
$
D
(A : A ) 2A : A ,
Dt
:A A:A
A
:A A:A
A
(A W A A W) : A A : (A W A A W)
$
A : A (W A ) : A ( A W ) : A A : A A : ( W A ) A : ( A W)
$
2A : A 2 A : (W A ) 2 A : ( A W) 2 A : A
282
where we have applied the commutative property of the double scalar product, i.e.
A : B B : A . Note that, due to the symmetry of A the following condition is satisfied:
A : (W A )
4.3.1.5
A ij (Wik A kj )
A jk ( A ji Wik )
A : (A W)
Let us now refer back to the Polar Decomposition of F , (see Chapter 2), in which we
obtained the following equation for the spin tensor:
1
U 1 U
1 U R T R R T
R U
2
(4.76)
W R R T , and by
l W
A
o A
(4.77)
WT A A W
A
WA AW
A
which is the same equation as that obtained in (4.67). Then, we can define the Green-Naghdi
rate, also known as the Polar rate or Green-McInnis rate, by:
4.3.2
A R R T
A A R R
(4.78)
The material time derivative of the first Piola-Kirchhoff stress tensor, (see Eq. (4.34)),
becomes:
Q P
P*
Ql
Then by substituting Q
we obtain:
P Q P Q l
Q
P *
P * l
*T
l
*T
l
Q P l
*T
P *
P Q P
Q
(4.79)
Q P Q P
Ql
P l * Q
P Q P
,
T
P*
(4.80)
Note that the orthogonal transformation of P obeys the vector transformation law so the
rate P l T P is objective because it has the same structure as the convective rate
presented in (4.59).
J F T J F T J F T
(4.81)
Additionally, if we take into account the material time derivative of the Jacobian
determinant, (see Chapter 2), we have:
J
J Tr ( l )
J Tr (D W )
J Tr (D)
J C 1 : E
J 1
C :C
2
JF T : F
(4.82)
283
P
F T
(4.83)
or:
>
P J Tr (D) l
@ F
T
(4.84)
* J * J * J Tr (D) * J *
Q T Q Q T Q Q
T
J Tr (D) * J Q
J * *
J * , so the
(4.85)
and Q
T
where we have substituted * into the equation in (4.57). Then by substituting Q
given in the equation (4.42), i.e. Q
obtain:
*
J
*
Tr (D)
, l
Q Q
Q Q l and
Q
QT Q l QT
QT l
T
Q
*T
l
Q Q T
Q QT l
Q l
QT
Q Tr (D) l l
*T
Q T , we
(4.86)
l
* * l *
*T
(4.87)
* * l *
Q Tr (D) l l
(4.88)
Hence, we obtain a new objective stress rate called the Truesdell stress rate:
7
l l
Tr (D)
(4.89)
Let us consider the Oldroyd rate of the Cauchy stress tensor l l T , and if
we use the Truesdell stress rate we can conclude that:
7
(4.90)
Tr (D)
We can also relate the Oldroyd rate of the Kirchhoff stress tensor l l
We can also prove the above equation is valid by starting from (4.87):
as:
(4.91)
284
*
l
*T
l
*T
* l
*T
*
J
* * l *
*
J
l *
* * l *
Q Tr (D) l l
Q
Q J Tr (D) l l Q
Q Tr (D) l l
(4.92)
thus,
T
Q T * l * * * l * Q
J Tr (D) l l
(4.93)
F 1 F T F 1 F T F 1 F T
F 1 l F T F 1 F T F 1 l
F
1
l l
F T
(4.94)
T
F 1 F T
J F 1 F T
(4.95)
J * we obtain:
D
* J * J * J * J (Q Q T ) J * J Q Q T Q Q T Q Q T (4.96)
Dt
W * Q Q W and J
Also if we consider the relationship Q
equation we obtain:
T
* J Tr (D) * J W * Q Q W Q T Q Q T Q W * Q Q W
(4.97)
* J Tr (D) * J W * Q Q T Q W Q T Q Q T
T
Q Q T W * Q W T Q T
* W * Q J Q T Q J Q T W *
J Tr (D) * J Q W Q T
Q Q T Q W T Q T
(4.98)
or
*
*
*
T
T
*
W
*
W
J Q >Tr (D) W W @ Q
T
Q QT
(4.99)
which give us the relationship between the Jaumann-Zaremba rate of the Kirchhoff stress
tensor and the Cauchy stress tensor:
(4.100)
5
The Fundamental Equations
of Continuum Mechanics
5.1 Introduction
The fundamental equations of continuum mechanics are based on the conservation
principles of certain physical quantities. We consider five of these to establish the basic
equations that govern the Initial Boundary Value Problem (IBVP), namely:
In this chapter we will address the fundamental principle of mechanics in the reference and
current configurations. At the end of the Chapter we will show that these principles are
insufficient to establish the IBVP set of partial differential equations, so, it is necessary to
add certain equations to fully resolve this problem. Then, we will introduce some concepts
and theorems to develop the concepts in this chapter.
5.2 Density
&
285
286
volume. Another density function we can quote is energy density, which measures stored
energy per unit volume. The term specific will be used to denote the amount of the
property per unit mass.
5.2.1
Mass Density
Any continuous medium is caused by a positive scalar quantity called mass. It is assumed
that the mass is continuously distributed throughout the continuum.
We will next review the concept of mass density introduced in Chapter 2. Let us consider a
sphere of infinitesimal radius centered at point P in the reference configuration, (see
Figure 5.1). The material contained in this sphere is denoted by 'm and the sphere volume
is represented by 'V0 . Then, the mass density S 0 , in the reference configuration, is
defined by the limit:
&
&
S ( x , t 0) S 0 ( X )
(t
0) { t 0
"im
'V0 o0
'm
'V0
dm
dV0
(5.1)
t
'V0 o 0
&
S0 ( X )
'
V
o
0
&
S ( x ,t )
x3
x1
x2
Reference configuration
Current configuration
Figure 5.1: Mass density.
S S ( x, t )
"im
'V o 0
&
&
'm
'V
dm
dV
(5.2)
Then the functions S ( x , t ) and S 0 ( X ) are continuous density functions and are
interrelated to each other, (see Chapter 2), by:
&
S0 (X )
&
JS ( x , t )
(5.3)
5.3 Flux
The properties conferred by density (e.g. mass, energy, entropy, etc.) are mobile and the rate
of change and direction of these quantities are assigned by the flux vector, usually denoted
& &
by q( x , t ) . With this information, we can define the amount of property that passes
through a differential area element da per unit time, (see Figure 5.2), as:
&
q n da
&
q cos D da q n da
(5.4)
287
&
&
where n is the unit normal vector, and D is the angle formed between q and qn . Note
&
&
that, only the normal vector qn crosses the surface, since the tangential vector q s remains
on the surface da . As an example of flux, we can mention the mass flux vector which is
represented by q S v . With regard to the SI unit we have: >q@
&
&
kg
where q represents
2
m s
&
&
&
J
where q refers to the energy flux vector.
2
m s
&
&
q
qn
D
da
Bt
x3
x2
x1
Let ) ( x , t ) be an Eulerian scalar field which describes a certain physical quantity per unit
&
volume. If ) ( x , t ) is continuous and differentiable, we can state that:
&
D
) ( x , t )dV
Dt V
&
&
&
&
dV Dt )( x, t ) )( x, t ) Dt (dV )
dV Dt )( x, t ) )( x, t )
&
x
v dV
&
&
&
Dt )( x, t ) )( x, t )
&
x
(5.5)
v dV
&
&
&
wv k
dV
k
Dt )( x, t ) )( x, t ) wx
(5.6)
Then by using the material time derivative operator we can still state that:
&
D
) ( x , t )dV
Dt V
w) ( x& , t )
&
& wv
w) ( x , t )
) ( x , t ) k dV
w
w
wx k
t
x
p
V
&
w) ( x , t )
&
w
) ( x , t ) v p dV
w
w
t
x
vp
(5.7)
This last equation is known as the Reynolds transport theorem and can be represented by the
following equations:
288
&
D
) ( x , t )dV
Dt V
&
&
Dt )( x, t ) )( x, t )
&
x
v dV
&
&
& &
w) ( x , t )
x& ) ( x , t ) v dV
wt
V
&
& &
w) ( x , t )
dV ) ( x , t )v n dS
w
t
V
S
V
(5.8)
where Vt is the control volume, S t is the control surface, and n is the outward unit
normal to the boundary S t of Bt . The first term on the right of equation is the local rate
of change of the property ) in the domain Vt , while the second term characterizes the
&
transport of )v , that leaves the domain Vt via the surface S t , (see Figure 5.3).
control volume
St
Vt
&
(Gv )
&
qn
&
wG( x , t )
wt
&
x
>(Gv& ) n @ n
control surface
Problem 5.1: Prove that Reynolds transport theorem is valid in the following equation:
D
) dV
Dt V
D
) JdV 0
Dt V
(5.9)
Solution:
D
) JdV 0
Dt V
5.4.1
V0
D)
DJ
)
dV 0
Dt
Dt
V0
Reynolds Transport
Discontinuities
&
D)
J) x& v dV 0
Dt
Theorem
for
D)
Dt
&
) x& v dV
Volumes
with
S
S
289
6(t )
B
6(t )
&
Z
n
6
6
n
&
x 6 (t )
(5.10)
Then, the unit normal vector n on the surface 6(t ) is given by the equation:
&
n ( x , t )
f 6
f 6
(5.11)
For material points belonging to the surface 6(t ) , the normal component of the velocity,
Z n , is defined as:
Zn
wf 6
wt
f 6
(5.12)
&
&
Z n n Z
wf 6
wt
f 6
f 6
f 6
&
wf 6
&
f 6 Z 0
wt
Df 6
Dt
(5.13)
Let A be a second-order tensor field and let us consider that A and A are the values of
A in the boundaries 6 and 6 , respectively. Then, we can define the jump of A as:
>>A @@
A A
(5.14)
Then by applying Gauss theorem (the divergence theorem) for the two domains B and
B we can obtain, respectively:
290
A dV A n dS A
n dS
A dV A n dS A
n dS
(5.15)
A dV
V V
A n dS A
S S
A dV
V V
n dS
n
n dS
(5.16)
>>A @@
A A the above
n and
A n dS >>A @@ n dS
S S
(5.17)
The Reynolds transport theorem can be modified for the case in which there is a singular
&
surface 6(t ) , which moves at the velocity Z , (see Figure 5.4). Then by applying the
equation in (5.8) to the two domains B and B , whose contours are S 6 and
S 6 , respectively, we obtain:
&
& &
& &
w) ( x , t )
dV ) ( x , t )v n dS ) ( x , t )Z n dS
w
t
V
S
6
&
& &
& &
w) ( x , t )
dV ) ( x , t )v n dS ) ( x , t )Z n dS
w
t
V
S
6
&
D
) ( x , t )dV
Dt V
&
D
) ( x , t )dV
Dt V
Then by adding the two equations above, and once again considering that n
and >>) @@ ) ) , we can conclude that:
&
D
) ( x , t )dV
Dt V V
V V
(5.18)
n
&
& &
&
w) ( x , t )
dV
) ( x , t )v n dS >>) @@ Z n dS
wt
6
S S
n
(5.19)
&
)( x, t )v n dS
S S
& &
& &
)( x, t )v dV >>)( x, t )v @@ n dS
(5.20)
&
x
V V
&
w) ( x , t )
x&
wt
V V
&
&
&
&
which results in the Reynolds transport theorem for domains with discontinuities:
&
D
) ( x , t )dV
Dt V 6
&
w) ( x , t )
x&
wt
V 6
)( x, t ) v dV >>) v Z@@ n dS
&
&
&
&
(5.22)
or
&
D
) ( x , t )dV
Dt V 6
&
&
&
& &
D) ( x , t )
) ( x , t ) x& v dV >>) v Z @@ n dS
Dt
V 6
6
291
&
w) ( x , t )
x&
wt
V
)( x, t ) v dV
&
&
(5.23)
z 0 source or sink
If the term on the left of the equation is nonzero this means that somewhere in the domain
there is a property source or sink, which can be represented locally by the variable Q .
Then, Q ! 0 indicates that there is a source, and Q 0 that there is a sink. For example, if
the property in question is mass density (mass per unit volume) in general Q 0 . However,
if there is a tumor (cells with abnormal growth) in a biological organism, we can establish a
law (at the macroscopic level) that indicates how the mass changes over time (source),
without regard to individual cells. Then, another example of a source we can cite is the
internal heat generated by a chemical reaction, such as in cement hydration. The effect of
the chemical reaction at the macroscopic level can be represented by a variable that
provides the amount of heat generated per unit volume and per unit time (the internal heat
source).
&
&
Note that the term ()v ) shows the flux of the property ) . Then, if ()v ) represents the
&
J
, and if we are dealing with mass
m2s
kg
. As we have seen before, in general, the flux is represented
m2s
by q , with which we can establish the local form (strong) of conservation law and which is
denoted by the following continuity equation:
Q
&
w) ( x , t )
x&
wt
& &
q( x, t ) Continuity equation
>) @
s
(5.24)
where >) @ is the SI unit of the physical quantity per unit volume.
V0
dV
S dV
>kg @
(5.25)
292
As a result of conservation of mass, the material time derivative of the total mass is zero,
i.e.
D
m 0 , then:
Dt
D
m
Dt
&
D
S ( x, t ) dV
Dt V
&
&
&
&
&
&
D
&
Dt >S ( x , t )@ S ( x , t ) x v dV
(5.26)
or in indicial notation:
D
wv k
dV
k
&
Dt >S( x, t )@ S wx
kg
s
(5.27)
If the above equation is valid for the entire domain, then it must also be satisfied locally:
DS
Sv k , k
Dt
kg
sm 3
(5.28)
which is the mass continuity equation in Eulerian description and is expressed in tensorial
notation as:
&
DS
S ( x& v ) 0
Dt
Dx
Dt
(5.29)
wx wx
v k , the mass
wt wx k
wv
wS
wS
vp
S k
wt
wx p
wx k
wS
w
( Sv k )
wt wx k
wS
( Sv k ) , k
wt
(5.30)
&
wS
The mass continuity equation
x& (S v ) 0
(Eulerian description)
wt
(5.31)
We could have obtained the same equation in (5.31) by means of Reynolds transport
&
&
theorem, i.e. in the equation (5.8) we substitute ) ( x , t ) for S ( x , t ) , which means:
&
D
S ( x, t )dV
Dt V
&
wS ( x , t )
x&
wt
V
S ( x, t ) v dV
&
&
&
wS ( x , t )
x&
wt
&
&
S ( x, t ) v
(5.32)
We could also have obtained the mass continuity equation in (5.29) by means of the
principle of conservation of mass in a differential volume element dx1 dx 2 dx 3 , (see Figure
5.5), in which the following is satisfied:
Mass
accumulation
Inward mass
flux
Outward mass
flux
The rate of mass entering through face A is represented by the mass flux (Sv1 ) x1 dx 2 dx3 ,
w (Sv1 )
while the rate of mass that goes through face B is given by Sv1
dx1 dx 2 dx3 .
x
w
1
293
Likewise, we can obtain the rate of change of mass in other faces. Moreover, by applying
the conservation of mass for the differential volume element we obtain:
dx1 dx 2 dx3
wS
wt
Sv1 dx 2 dx3 Sv 2 dx1 dx3 Sv3 dx1 dx 2 Sv1 w(Sv1 ) dx1 dx 2 dx3
wx1
w ( Sv 2 )
w ( Sv 3 )
dx 2 dx1 dx 3 Sv3
dx3 dx1 dx 2
Sv 2
wx 2
wx3
wS
wt
(5.33)
and
wS wS
vi
wt wx i
wv
wv
wv
S 1 2 3
x
x
w
w
wx3
2
1
wvi
&
DS
S x& v 0
xi
Dt
w
& &
S
(5.34)
v Tr (v )
DS
Dt
face A
x2
face B
w (Sv1 )
dx1 dx 2 dx 3
Sv1
x
w
1
dx 2
Sv1 dx 2 dx 3
dx3
x1
dx1
x1
x3
5.6.1
The mass continuity equation in (5.29) can also be expressed in Lagrangian description
(material). To do this, we can start from the conservation of mass which establishes:
V0
&
0 ( X ) dV 0
&
&
( x , t ) J dV
S( x, t ) dV S
&
V0
f (X )
(5.35)
Since the above equation is valid for any volume it means that it will be valid locally too,
i.e.:
&
S0 (X ) J S
(5.36)
294
D
The mass continuity equation
(S J ) 0
(Lagrangian description)
Dt
&
&
DPij ( x , t )
(5.38)
dV
Dt
(5.37)
where Pij ( x , t ) is a continuum property per unit mass, which can be a scalar, a vector or
higher order tensor.
Solution: It was proven in equation (5.6) that:
D
&
D
) ( x , t )dV
Dt V
&
wv p
dV
p
&
Dt ) ( x, t ) ) ( x, t ) wx
D
S Pij dV
Dt V
Dt (SP
ij )
SPij
D
Pij Pij
S
Dt
wv p
dV
wx p
S Dt P
Pij
ij
DS
wv
S k
wx k
Dt
wv
DS
SPij k dV
wx k
Dt
dV
0
mass continuity equation
DPij
dV
Dt
Sa
&
& &
w
(S v ) x& (S v v )
wt
(5.39)
wt
Dt
ai
w (S v i )
dV (S v i v k )n k dS
wt
S
(S v i v k ) ,k dV
dV (S v i v k ) ,k dV
S ai dV
V
wt
wt
wt
&
Sa
&
& &
w (S v )
x& (S v v )
wt
5.6.2
295
Incompressibility
DS
Dt
&
S x& v Sv k ,k
(5.40)
thus
&
x& v
(5.41)
0 ;
DS
{ S
Dt
0 ; S
S0 ; J 1
(5.42)
or
v k ,k
wv1 wv 2 wv3
wx1 wx 2 wx3
Tr ( l )
Tr (D) 0
(5.43)
where l denotes the spatial velocity gradient, and D is the rate-of-deformation tensor
which is equal to the symmetrical part of l , (see Chapter 2).
5.6.3
Now, let us consider a domain where there is a singular surface 6(t ) as established in
subsection 5.5.1, (see Figure 5.4). Based on the conservation of mass we have:
D
SdV
Dt V
(5.44)
and if we consider the Reynolds transport theorem with discontinuities, (see equation
(5.22)), in which ) S , we obtain:
DS
&
& &
S x& v dV >>S v Z @@ n dS 0
(5.45)
V
6
&
& &
where the mass density S ( x , t ) , and the velocity v ( x , t ) are continuous differentiable
& &
functions in V 6 , and >>S v Z @@ is also a continuous differentiable function on 6 . The
D
SdV
Dt V 6
6 Dt
global balance law is valid for any arbitrary parts of the volume and for the discontinuous
surface, hence it holds that:
&
DS
S x& v 0 in V 6
Dt
>>S v& Z& @@ n 0 on 6
(5.46)
296
vi
for t t 0
{
S k
wx k
Dt
Dt
wx i
Thus,
dS
dt
1 t wx i
3S
1 t
3
1 t
1 t
dS
wx k
dt
3dt
S
1 t
Then by integrating the both sides of the above equation we obtain:
dS
3dt
S 1 t lnS 3 ln(1 t ) C
The constant of integration C is obtained by means of the above equation if we refer to
&
the initial condition t 0 , in which S ( x , t 0) S 0 , thus
ln S 0 3 ln(1 0) C C lnS 0
ln S
1
ln
3
(1 t )
3 ln(1 t ) ln S 0
ln S 0
S0
ln
3
(1 t )
S0
1 t 3
dx i
dt
xi
1 t
dx i
xi
dt
1 t
dx i
xi
dt
1 t
ln(1 t ) K i
lnx i
(5.47)
X i , we obtain:
ln(1 0) K i
K i lnX i
Additionally, by substituting the value of K i into the equation (5.47) we obtain:
lnx i ln(1 t ) ln X i
ln( x i ) ln> X i (1 t ) @
Hence we can conclude that x i X i (1 t ) , which gives us x1 X 1 (1 t ) , x 2 X 2 (1 t ) ,
lnX i
x3
1 t 3
S
1
t
1
t
1
t S0
x1
X1
x2
X2
x3
X3
S0
, we obtain:
S x1 x 2 x 3
S0 X1 X 2 X 3
297
Linear Momentum
Let us consider the body, Bt , in motion which is subjected both to body forces (per unit
& &
&
&
mass), b( x , t ) , and to surface forces, t * ( x , t ) , acting on the surface S , (see Figure 5.6).
& &
Let v ( x , t ) be the Eulerian velocity field, then we can define the linear momentum of the
mass system Bt as:
&
L
&
&
v dm S v dV
V
Bt
& &
Sb( x , t )
&
x
x3
& &
Sv ( x , t )
(5.48)
& &
t * ( x , n , t )
Bt
dV
kg m
s
Linear momentum
x1
x2
5.7.2
The principle of conservation of linear momentum, based on Newtons second law, states
that the rate of change of the linear momentum of an arbitrary part of a continuous
medium is equal to the resultant force (body and surface forces) acting on the part in
question, then:
&*
&
dS S b dV
&
D
S v dV
Dt V
t dS S b dV
*
i
D
S v i dV
Dt V
(5.49)
The equation in (5.49) represents the global form of the principle of conservation of linear
momentum and by applying t *i V ij n j we obtain:
ij n j dS
Guass '
Theorem
Vij , j dV
thus,
Sb i dV
V
D
Svi dV
Dt V
Uvi dV
(5.50)
298
(V
ij , j
Sb i Svi )dV
kg m
s2
0i
(5.51)
Pa
m
(5.52)
If the above equation is valid for the entire volume, it is also valid locally, i.e.:
V ij , j Sb i Svi
kg
s2 m2
0i
N
m3
which are known as the equations of motion or Cauchys first equation of motion:
&
&
&
x& Sb Sv Sa
(5.53)
S0
0 dS 0
&
S 0 b 0 dV0
V0
S A J dV
&
dS S b dV
P N
0
0
0
0
&
D
S 0V dV0
Dt V
V0
&
(5.54)
&
S 0 A dV0
V0
& &
V0
S0
&
&
&
D
S V J dV0
Dt V
& &
zN
P , since
forces vector per unit mass in the undeformed configuration. Note that P N
P is a non-symmetric tensor. Then by applying Gauss theorem (the divergence theorem)
to the surface integral we obtain:
&
X
&
&
0
&
A dV0
P S 0 b0 S 0
V0
(5.55)
Then, the local form of the equations of motion in material description (Lagrangian) can be
expressed as:
&
X& P S 0 b 0
X&
5.7.2.1
&
S0 A
&
&
F S S 0 b 0 S 0 A (Lagrangian description)
(5.56)
In the exceptional cases when we have a static or quasi-static equilibrium, the acceleration
components are zero, thus we obtain the equilibrium equations as:
& &
x& Sb 0
(5.57)
wx 2
wx3
wx1
wV 21 wV 22 wV 23
Sb 2
wx 2
wx 3
wx1
wV 31 wV 32 wV 33
wx wx wx Sb 3
2
3
1
0
0
0
(5.58)
299
Then by using both the engineering and Voigt notation, the equilibrium equations can be
expressed as follows:
V
w
w
w x
0
0
0
x
wy
wz V y Sb
w
1
V z
w
w
w
0
0
0 Sb 2
W xy
wy
wx wz
w
w
w W Sb 3
yz
0
0
0
wz
wy wx W
xz
0
0 >L@T ^T` ^M `
0
^0`
(5.59)
>L @T
Additionally, the equilibrium equations in the Lagrangian description are given by:
&
X& P S 0 b 0
&
0
&
X& ( F S ) S 0 b 0
(5.60)
Problem 5.5: Find the equilibrium equations in engineering notation by means of the
differential volume element equilibrium ( dxdydz ). For this purpose consider that the
Cauchy stress tensor field in the differential volume element varies as indicated in Figure
5.7.
z
Rear face
Vz
Rear face
W xy
wV z
dz
wz
V yz
W xz
wV yz
wW xz
dz
wz
W xy
Vy
Vx
bx
W xy
wV x
dx
wx
wW yz
wy
wx
dz
dy
Vy
W xy
wW xy
W xz
dz
by
wW xz
dx
wx
W yz
wz
W yz
bz
W xz
Vx
wW xy
wy
wV y
wy
dy
dy
dx
dx
W xz
x
W yz
Rear face
Vz
dy
300
Solution:
To obtain the equilibrium equations we apply the force equilibrium condition in the
volume element. First, we evaluate the equilibrium force according to the x -direction:
wW xy
wV x
dx dydz V x dydz W xy
dy dxdz
wx
wy
wW
Sb x dxdydz V x
wW xy
wW
wV x
dxdydz
dxdydz xz dxdydz
wz
wx
wy
wW xy wW xz
wV
Sb x x
0
wz
wx
wy
wy
W yz dxdy W xy
wW yz
dy dxdz V y dxdz W yz
dz dxdy
wz
wW xy
dx dydz W xy dydz 0
wx
Sb y dxdydz V 22
Fy
wW xy
wx
wV y
wy
wW yz
wx z
Fz
0 , is given by:
wW
wV z
w
W
yz
W xz dzdy W yz
dy dxdz W yz dxdz 0
w
y
Sb z dxdydz V z
wW xz wW yz wV z
wz
wx
wy
wy
wz
wx
wW xy wV y wW yz
Sb y
wy
wx z
wx
wW
wW
wV z
xz yz
Sb z
wy
wz
wx
0
0
0
Problem 5.6: Let be the Cauchy stress tensor field, which is represented by its
components in the Cartesian basis as:
V11
x12 ;
V 22
V12
V 21
2 x1 x 2 ;
x 22 ;
x12 x 22
V 33
V 23
301
V 32
V 31
V13
Considering that the body is in equilibrium, find the body forces acting on the continuum.
& &
Solution: By applying the equilibrium equations, x& Sb 0 , we obtain:
V ij , j Sb i
0i
wV 11 wV12 wV13
Sb 1 0
wx 2
wx 3
wx1
wV 21 wV 22 wV 23
Sb 2 0
wx 2
wx 3
wx1
wV 31 wV 32 wV 33
Sb 3 0
wx 2
wx 3
wx1
2 x1 2 x1 Sb 1 0
2 x 2 2 x 2 Sb 2 0
Sb
3 0
Thus, to satisfy the equilibrium equations the following condition must be met:
4 x1 Sb 1 Sb 1 4 x1
4 x 2 S b 2 S b 2 4 x 2
Sb 3
&
Sb
4( x1 e 1 x 2 e 2 )
Problem 5.7: The equations of motion of a body are given, in Lagrangian description, by:
x1 X 1 BtX 3
x 2 X 2 BtX 3
x
3 X 3 Bt ( X 1 X 2 )
where B is a constant scalar. Find the mass density in the current configuration (S ) in
terms of the mass density of the reference configuration (S 0 ) , i.e. S S (S 0 ) .
Solution:
We can apply the equation S 0 JS , where J is the Jacobian determinant and is given by:
5.7.3
wx i
wX j
Thus, we obtain S
wx1
wX 1
wx 2
wX 1
wx 3
wX 1
S0
S0
1 2(Bt ) 2
wx1
wX 2
wx 2
wX 2
wx 3
wX 2
wx1
wX 3
wx 2
wX 3
wx 3
wX 3
Bt
Bt
Bt
Bt
1 2(Bt ) 2
Let us consider again a domain with a singular surface 6(t ) such as that discussed in
subsection 5.5.1, (see Figure 5.4). Then, the principle of conservation of linear momentum
becomes:
&
D
S v dV
Dt V 6
&
n dS S b dV
S 6
(5.61)
V 6
Then by applying the divergence theorem with discontinuities, (see Eq. (5.17)), we obtain:
&
D
S v dV
Dt V 6
V 6
&
x
S b dS >> @@ n
&
dV
(5.62)
302
&
&
&
&
&
& &
D ( Sv )
Sv x& v dV >>Sv v Z@@ n dS
Dt
V 6
6
(5.63)
Then by combining the above equation with the equation in (5.62) and by considering that
&
&
D ( Sv ) & D ( S )
D (v )
S
we obtain:
v
Dt
Dt
Dt
&
&
& D (S )
& D (v )
&
& &
v
S x& v S
x& S b dV >>Sv v Z @@ n dS
Dt
Dt
V 6
6
&
D (S )
S x& v
Dt
&
0
(5.64)
becomes:
&
&
&
& &
D (v )
x& S b dV >>Sv v Z @@ n dS
Dt
6 S
V
&
0
(5.65)
(5.66)
n n
(5.67)
(Eulerian description)
Angular Momentum
Once again let us consider Figure 5.6, and we can define the angular momentum of a mass
system with respect to the origin by:
&
HO
&
&
( x Sv ) dV
H O i (t )
(.
ijk
x j S v k ) dV
Angular momentum
&
>& @
The SI unit of H O is H O
&
kg m 2
, and H O
kg m 2
s2
Nm
J.
(5.68)
5.8.2
303
The principle of conservation of angular momentum states that the rate of change of
angular momentum with respect to a point is equal to the resultant moment (with respect
to this point) produced by all forces acting on the body under consideration.
Then by obtaining the resultant momentum with respect to the origin, (see Figure 5.6), and
by applying the principle of angular momentum, we obtain:
&*
&
(x t
&
&
)dS ( x Sb)dV
&
&
D
( x S v )dV
Dt V
>Nm@
(5.69)
NOTE: The equation in (5.69) is valid for those continuous media in which the forces
between particles are equal, opposite and collinear, and without any distributed moments.
The equation in (5.69) can be rewritten in indicial notation as:
(.
ijk
D
(. ijk x j S v k )dV
Dt V
S (.
S Dt (.
ijk
x j v k )dV
(5.70)
x v . ijk x j v k )dV
,j k
v
j
ijk
0i
Then by substituting t *k
theorem, we obtain:
( x j V kl ) ,l dV (. ijk x j Sb k )dV
ijk
(.
V
ijk
(.
ijk
ijk V kj
(5.71)
(.
(5.72)
ijk
x j Sa k )dV
E jl
x j Sa k )dV
. ijk x j (V kl ,l Sbk Sa k ) dV
0
0i
k
Equations of motion
. ijk V kj
. ijk V kj dV
(5.73)
0i
(5.74)
V jk
0i
V kj
Thus obtaining Cauchys second law of motion, also known as the Boltzmann postulate, the
symmetry of the Cauchy stress tensor is:
(5.75)
Then bearing in mind the relationship J 1P F T , the Boltzmann postulate in the
reference configuration becomes:
1
PFT
J
1
T
P
F
PFT
F PT
(5.76)
304
F S F T
F (F S)T
F S F T
F ST F T
(5.77)
Thus
(5.78)
S ST
Problem 5.8: Find the linear and angular momentum for a solid subjected to rigid body
motion.
x3c
&
F(n )
&
F( 2 )
Rigid body
Bt
&
x
x3
x2c
&
v
x1c
G - mass center
&
F(1)
x2
x1
Solution: According to Problem 2.16 in Chapter 2, we obtained the velocity for rigid body
motion as:
& & &
& &
v c ( x c )
&
where is the axial vector (angular velocity) associated with the antisymmetric tensor W
S c ( x c) dV S c dV S x dV S c dV
&
&
&
&
&
&
By definition
&
&
&
&
& &
c S dV S x dV c S dV
V
&
&
&
mx is the first moment of inertia, where m is the total mass, and
&
S x dV
&
x k is the vector position of the center of mass G . The first moment of inertia is equal to
& &
&
zero if the Cartesian system originates at the center of mass, so, S x c dV mx c 0 .
&
L
&
&
&
&
>
& &
& &
m c ( x c )
&
mv
&
( x Sv ) dV >x S c ( x c) @
&
Thus
&
&
&
&
&
&
dV
&
HO
&
&
&
&
&
&
305
&
&
S x c dV S x ( x ) dV S x ( c) dV
(5.79)
&
&
&
& &
&
& &
S x dV c S x ( x ) dV S x dV ( c )
V
&
&
&
which we obtain:
S >x k x k Zi x p Z p xi @ dV
S >x
S >x
S x ( x) dV S >( x x) ( x ) x @ dV , with
k
S >x
x k Z p E pi x p Z p x i dV
x k E pi x p x i Z p dV
x k E pi x p x i dV Z p
I O ip Z p
or in tensorial notation:
&
&
&
& &
&
&
&
&
IO
S x ( x ) dV S >( x x ) 1 ( x x )@ dV
where I O
& &
& &
S >( x x ) 1 ( x x )@ dV is the inertia tensor with respect to the origin O . As
S >( x x
1 1
I O 22
S >x
2
1
I O 12
S >( x x
1 1
I O 33
S >x
2
1
2
2
x 32 dV
x 22 dV
x 2 x 2 x 3 x 3 )E 12 x1 x 2 @ dV
S >x1 x 3 @ dV
V
I O 13
S >x
x 2 x 2 x 3 x 3 )E 11 x1 x1 @ dV
x 32 dV
S >x
S >x1 x 2 @ dV
I O 12
I O 13
I O 23
S >x 2 x 3 @ dV
I O 23
where I O 11 , I O 22 , I O 33 , are moments of inertia of the body relative to the reference point O ,
and I O 12 , I O 13 , I O 23 , are the products of inertia of the body relative to the reference point
O.
Returning to the equation in (5.79) we can state that:
&
&
&
& &
&
& &
S x dV c S x ( x ) dV S x dV ( c )
V
V
& &
&
& &
&
& &
& &
&
m x c I O m x ( c ) m x c ( c ) I O
& & &
Then by adding and subtracting the term m x x in the above equation we obtain:
& & & &
& & &
& &
&
&
&
& &
&
H O m x c c I O m x c ( x c ) m x ( x ) I O
& &
& &
& & &
& &
& &
& &
&
&
m x v m (x x) 1 ( x x) IO m x v m ( x x) (x x) 1 IO
& &
&
m x v I
& & &
m x v HG
&
HO
>
>
>
^>
306
>&
&
& &
I11 m( x 22 x32 ) ; I O 12
I 22 m( x12 x32 ) ; I O 23
I12 m( x1 x 2 )
I 23 m( x 2 x 3 )
I O 33
I33 m( x12 x 22 ) ;
I13 m( x1 x3 )
I O 13
Note that, the above equations represent the parallel axis theorem (Steiners theorem) from
Classical Mechanics.
Problem 5.9: Obtain the principle of conservation of linear momentum and angular
momentum for a solid subjected to rigid body motion.
Solution: We can start from the definition of the principle of conservation of linear
momentum which states that:
&
&
D
S v dV
Dt V
&
L
&
&
&
&
L mv
&
D
S v dV
Dt V
Then we have:
&
&
ma
&
ma
Now let us consider the principle of conservation of angular momentum which states:
&
M
By which we obtain:
&
&
D
( x S v )dV
Dt V
&
&
HO
or
&
&
D &
HO { HO
Dt
&
&
HG
where the equation of angular momentum H O was obtained in Problem 5.8. The set of
equations
equivalent:
&
&
ma
and
&
&
H G inform us that the following systems are
&
HG
&
F( 2 )
&
F(n )
G
=
&
F(1)
&
ma
G - center of mass
307
(5.80)
The energy supplied to, or removed from, the system per unit time can be any of three
kinds: thermal; chemical; or electromagnetic energy. In this publication we only consider
thermal energy as the energy added to the system. In such circumstances, the principle of
conservation of energy is known as the first law of thermodynamics. Mathematically, the
principle of conservation of energy, for continuum thermodynamics, is given by:
DK DU
Dt
Dt
DW DQ
Dt
Dt
J
s
(5.81)
where K is the kinetic energy, U is the internal energy, W is the work done by the
system, and Q is the energy added to the system.
Next, we will introduce the types of energy involved in the energy equation.
5.9.1
Kinetic Energy
The kinetic energy of the system represented in Figure 5.6 is given by:
K (t )
& &
1
S (v v )dV
2V
1
S (vi vi )dV
2V
kg m m
dV
3
s s
>J @
Kinetic energy
Nm
dV
Nm
(5.82)
J.
D
Dt
S (vi vi )dV
2 V
D
1
S (vi v i )dV
2 V Dt
1
S (vi vi vi vi )dV
2V
(5.83)
Thus
D
K(t ) { K
Dt
5.9.2
S v v
i i
dV
(5.84)
v (V
i
ij , j
Sb i )dV
308
(v i V ij ) , j
v i , j V ij vi V ij , j
v i V ij , j
(vi V ij ) , j vi , j V ij
,
(5.86)
l ij
&
where x& v { l is the spatial velocity gradient, which can be broken down into a
symmetric and an antisymmetric part, i.e. l D W , (see Chapter 2), where D is the rateof-deformation tensor and W is the spin tensor. The components of these tensors can be
expressed in terms of Eulerian velocity as:
l ij
1 wvi wv j 1 wv i wv j
2 wx j wxi 2 wx j wxi
vi, j
Dij
m
m s
Wij
(5.87)
Returning to the equation in (5.85), and considering the relationships in (5.86) and (5.87),
the rate of change of the kinetic energy becomes:
>v V
K
>(v V
v i , j V ij Sb i vi dV
ij , j
Sb v dV v V
i i
ij , j
ij ) , j
V ij (D ij Wij ) Sb i vi dV
(5.88)
dV V ij D ij dV
where we have taken into account that the double scalar product of a symmetric and
antisymmetric tensor is equal to zero, i.e. V ij Wij 0 or : W 0 . Then by applying the
divergence theorem to the second integral of the right side of the equation in (5.88), we
find that:
v V
i
ij , j
v V
dV
v t dS
*
i i
ij n j dS
(5.89)
By combining the above relationship with the equation in (5.88), we can still express the
rate of change of the kinetic energy as:
K
Sb v dV v t
*
i i
i i
dS
V ij D ij dV
V
Pint ( t )
Internal Mechanical Power
Pext (t )
External Mechanical Power
D
K
Dt
Pext (t ) Pint (t )
(5.90)
or
D
K Pint (t ) Pext (t )
Dt
(5.91)
where we have introduced the external mechanical power Pext (t ) , which is the rate of change
DW
, as:
of the work done by the external forces
Dt
Pext (t )
Pext (t )
& &
& &
t * v dS S b v dV
t *i vi dS
S b i vi dV
(5.92)
and the internal mechanical power, also known as the stress power, which is the rate of change of
the work done by the internal forces:
Pint
V D
ij
ij dV
: D dV Tr
D dV
Tr D dV
(5.93)
309
NOTE: The SI unit of power is the watt, W { J / s , i.e. one joule ( J ) per second ( s ),
m
N m
dV
2
ms
Pa m s dV m
J
s
dV
W.
We can also define the stress power per unit volume, denoted by w int (t ) , as:
Tr ( D)
w int (t ) : D
(5.94)
Then by starting from the stress power in the current configuration we can also express the
stress power as a function of the other stress tensors, i.e.:
J : D dV
: D dV ,
: D dV
V0
(5.95)
V0
: D dV
V0
dV0
F 1 F jk dV0
lj
dV0
1
1
ik F jk F pi E pl Flj
ik F jk D ij
V0
V0
dV0
ik F jk
1
ik F pi E pk
1
pi Pik E pk
dV0
V0
pk E pk
ik l ij F jk
P : F dV
dV
0
ik Fik
V0
V0
0
(5.96)
dV0
V0
dV0
S : E dV
( l ij Wij ) dV0
V0
Glk
V0
V0
F pi1 E pl Flj1 , we
V0
1
ik F pi E pl
F PT
1
S : C dV0
2V
J P : F dV S
V0
P : F dV
which proves that the rate of change of the deformation gradient and the first PiolaKirchhoff stress tensor are conjugate quantities ( P : F ). Other conjugate quantities are: the
second Piola-Kirchhoff stress tensor and the rate of change of the Green-Lagrange strain
tensor ( S : E ); the Kirchhoff stress tensor and the rate-of-deformation tensor ( : D ).
Furthermore, we can show that T : U is already a conjugate pair. To prove this, let us
consider the relationship P R T , where T U S is the Biot stress tensor, and R is the
orthogonal tensor from the polar decomposition, and U UT is the right stretch tensor.
Then if we refer to the right polar decomposition, i.e. F R U F R U R U , we
obtain:
P : F
R T : R U R U
R T : R U R T : R U
T UT : R T R T : U
U S UT : R T R T : U
T :U
Pij Fij
ip T pj
R
ik U kj
R ik U kj
R ip T pj R ik U kj R ip T pj R ik U kj
( T U )(R R ) T U
pj
kj
ip
ik
kj
kj
(U pq S qj U kj )(R ip R ik ) Tkj U kj
T U
kj
kj
(5.97)
310
Note that U S U : R T R
symmetrical and R
becomes:
R
U S UT
U S U is
T :H
T :U
(5.98)
where H U 1 is the Biot strain tensor, (see Chapter 2) and if we know that U is
symmetrical, it is also possible to express the above relationship as:
T sym : U
( T sym T skew ) : U
P : F
(5.99)
Then, if we take into account all the equations obtained before, we can summarize the
stress power per unit volume by:
w int
:D
S
P : F
S0
1
S : C
2
: D S : E P : F
w int
5.9.3
1
P : F
J
T :H
T :U
(5.100)
(5.101)
If we compare the equation given in (5.91) with the energy equation (5.81), i.e.:
D
K Pint (t ) Pext (t )
Dt
DK DU
Dt
Dt
DW DQ
Dt
Dt
,
(5.102)
we can observe that the equation in (5.91) is an exceptional case of the energy equation
where only mechanical energy is considered. In this case the principle of conservation of
energy is known as the balance of mechanical energy which is otherwise known as the theorem of
power extended:
D 1 2
Sv dV V ij D ij dV
Dt V 2
V
D 1 2
Sv dV : DdV
Dt V 2
V
Rate of change
of the Kinetic energy
Stress power
Sb v dV v t dS
*
i i
i i
& &
Sb v dV
&* &
v dS
Balance of
mechanical energy
S
External mechanical power
OBS.: In rigid body motion D 0 is satisfied, so, the stress power (internal
mechanical power) is zero Pint (t ) 0 , then it holds that K Pext (t ) .
If K is discarded, which characterizes a static or quasi-static regime, it holds that
Pint (t ) Pext (t ) .
(5.103)
311
Problem 5.10: Find the kinetic energy related to rigid body motion in terms of the inertia
tensor, (see Problem 5.8 and Problem 5.9).
& & &
& &
Solution: The rigid body motion velocity can be expressed as v c ( x c ) . Then, the
kinetic energy becomes:
& &
1
S (v v )dV
2V
>
@>
1 & &
&
& &
S c ( x& c& ) c&
( x c) dV
2V
&
& & &
Using the following vector sum x x x c , where x is the mass center vector position,
&
and xc is the particle vector position with respect to the system that has its origin in the
K (t )
&
Note that v
or:
^>
@>
^>
@`
& &
& &
& &
& & &
&
1
S c (( x x c) c) c (( x x c) c ) dV
2V
& &
& &
& &
& &
& &
& &
1
S c ( x c ) ( x c) c ( x c ) ( x c) dV
2V
& &
& &
c ( x c ) is the center of mass velocity, thus:
&
&
& &
& &
1
K (t )
S v ( x c) v ( x c) dV
2V
K (t )
@ >
^>
@>
@`
@`
& &
& & &
& & &
& &
& &
1
1
1
1
Sv v dV
Sv ( x c) dV
S ( x c) v dV S ( x c) ( x c) dV
2V
2V
2V
2V
K (t )
& &
& & &
& &
& &
1
1
Sv v dV Sv ( x c) dV
S ( x c) ( x c) dV
2
2V
V
V
& &
1
Sv v dV
2V
1 &
v
2
1
mv 2
2
S dV
& &
& &
&
&
2)
v Sx c dV v ( m ,
x c) 0
&
V
V
0
&
Note that, the system xc is located at the center of mass ( G ), hence the center of mass
&
vector position related to the system xc is zero.
& &
& &
3) S ( x c) ( x c) dV
&
&
&
&
S >( x c) ( x c)@ dV S .
ijk Z j x kc
. ipq Z p x qc dV
S (E
E kq E jq E kp )Z j x kc Z p x qc dV
S Z
(E jp E kq x kc x qc E jq E kp x kc x qc )Z p dV
S Z
(E jp x kc x kc x cp x cj )Z p dV
Z j S (E jp x kc x kc x cj x cp ) dV Z p
Z j I jp Z p
jp
312
&
&
&
&
&
& &
&
&
S >( x c x c) 1 ( x c x c) @ dV
V
&
&
I
&
S >( x c) ( x c)@ dV
where I is the inertia pseudo-tensor related to the system located at the center of mass,
(see Problem 5.8).
Then if we bear in mind all the above considerations, the kinetic energy equation for rigid
body motion becomes:
K (t )
& &
& & &
& &
& &
1
1
1
Sv v dV
2Sv ( x c) dV
S ( x c) ( x c) dV
2V
2
2V
V
&
1&
1
mv 2 I
2
2
K (t )
Iij
2
2
S x 2c x 3c dV
V
S >x c x c @ dV
1 2
V
S >x1c x 3c @ dV
V
>
S >x1c x 2c @ dV
>
S x1c x3c dV
S >x 2c x 3c @ dV
S >x1c x 3c @ dV
V
S >x 2c x 3c @ dV
V
S x1c 2 x 2c 2 dV
V
>
I11
I12
I
13
I12
I 22
I 23
I13
I 23
I 33
1
1
mv 2 Z k I kj Z j
2
2
1
1
mv 2 >Z1
2
2
Z2
>
I11
Z 3 @ I12
I
13
I12
I 22
I 23
I13 Z1
I 23 Z 2
I 33 Z 3
1
1
mv 2 I11Z12 I 22 Z 22 I 33 Z32 2 I12 Z1 Z 2 2 I13 Z1 Z3 2 I 23 Z 2 Z3
2
2
K (t )
5.9.4
>
1
1
mv 2 I11Z12 I 22 Z 22 I 33 Z32 2 I12 Z1 Z 2 2 I13 Z1Z 3 2 I 23 Z 2 Z 3
2
2
If we take a handful of atoms (the material point) and we evaluate the average of all forms
of energy present in it we obtain what is known as the internal energy. Continuum
thermodynamics usually presents the rate of change of the internal energy as:
DU
Dt
D
Su dV
Dt V
J
s
Su dV
J
. For example, for an
kg
where u is the specific internal energy, i.e. energy per unit mass, >u @
ideal gas the specific internal energy is given by u c v T
(5.104)
p
c p T , where T is the
, S
I
temperature, c v is the specific heat capacity at a constant volume, I is the specific entropy,
313
energy is given by
5.9.5
DU
Dt
: D dV .
Thermal Power
We define thermal power as the rate of increase of total heat in the continuum, which is
denoted by
DQ
. The contribution of thermal power considered here is caused by:
Dt
The Cauchy heat flux (non convective, i.e. without mass transport);
q( x, t ) n dS
J
s
(5.105)
S r dV
J
s
(5.106)
&
where r ( x , t ) is the radiant heat constant (also called the heat source) per unit mass per unit
&
&
J
, and S ( x , t ) is the mass density.
s kg
Then by considering the heat flux (incoming) and the heat source, we can define the thermal
power (the rate of thermal work) as:
DQ
Dt
&
S r dV q n dS
J
s
(5.107)
314
& &
&
q n ( x , t ) (q n ) n q n n
Current configuration
& &
q( x , t )
Bt
x3
dV
dS
&
>q& @
Sr ( x , t )
x2
>r @
x1
J
m2s
J
kg s
5.9.6
Once we know what forms of energy are involved in a system we can provide the energy
equation by starting from that in (5.81):
DK DU
Dt
Dt
DW DQ
Dt
Dt
(5.108)
), but there is
The mechanical power and the thermal power are not exact differentials ( Dx
Dt
experimental evidence showing that the sum of mechanical and thermal power is already an
exact differential, (Mase(1977)).
Considering only the mechanical and thermal energy, the principle of conservation energy
becomes what is known as the first law of thermodynamics, which postulates the
interchangeability of mechanical and thermal energy. Then, the equation in (5.108)
becomes:
Vijn j
vv
D
S i i dV S udV
2
Dt V
V
t *i vi dS Svi b i dV S rdV q i n i dS
(5.109)
Sq
Then by using divergence theorem to transform the surface integral into the volume
integral we obtain:
vv
D
S i i dV SudV
2
Dt V
V
Sv v dV SudV
i i
V v dV Sv b dV S rdV q
>V v Sv b S r q @ dV
ij i , j
ij i , j
i ,i
dV
(5.110)
i ,i
SudV V
ij , j v i
V ij v i , j Sb i vi S r q i ,i Svi vi dV
V
SudV v V S
bSv
i
ij , j
0i
ij v i , j
Sr q i ,i dV
(5.111)
Su dV >V D
ij
ij
315
Wij Sr q i ,i dV
Su dV V D
ij
ij
(5.112)
J
m3 s
W
(5.113)
m 3
Sr q i ,i dV
The local form of the above equation is known as the energy equation:
Su V ij D ij Sr q i ,i
which is expressed in tensorial notation as:
&
(5.114)
NOTE: For a purely mechanical problem in which there is no internal heat production
& &
( r 0 ) nor heat flux q 0 , the energy equation becomes:
u
s kg
:D
m3 N m
kg m 2 m s
5.9.6.1
Nm
s kg
(5.115)
s kg
The energy equation (5.114) can also be established in Lagrangian description (material
description). From the equation in (5.112), the integral related to the integral energy can be
written in the reference configuration as:
&
&
V0
&
&
( X , t )dV0
0 ( X )u
(5.116)
V0
The integral associated with stress power can be established in the reference and current
configuration, (see equations (5.100) and (5.101)), as shown bellow:
S
J : D dV
: DdV ,
: DdV S : E dV 2 S : CdV P : FdV S
V0
V0
V0
V0
V0
P : F dV
(5.117)
Similarly for the integral related with the heat source, i.e.:
&
&
S( x, t ) r ( x, t ) dV JS r dV S
V
current configuration
&
0 (X )
&
r ( X , t ) dV0
V0
V0
(5.118)
reference configuration
Finally, we can address the integral related to the heat flux. The amount of heat that passes
through the area element da in the current configuration must in theory be the same as,
that which passes through the area element dA in the reference configuration, (see Figure
5.9). Then the following relationship must be met:
& & &
&
q 0 dA q da
(5.119)
&
where q 0 is the heat flux in the reference configuration. Then if we use Nansons formula
&
&
da JF T dA , obtained in Chapter 2, the equation in (5.119) becomes:
316
&
&
q 0 dA
&
&
&
J q F T dA q 0
&
q0
&
&
J q F T q
&
J 1 q 0 F T
&
q
F
&
dA
(5.120)
&
da
&
&
da JF T dA
&
&
q J 1 q 0 F T
Reference configuration
Current configuration
Figure 5.9: Heat flux.
&
x& qdV can be written in the reference configuration as:
&
x
&
qdV
i ,i
wq i
J wx
dV
V0
w 1
q 0 k Fik dV0
i J
J wx
dV0
V0
wq 1
w 1
J 0 k Fik J q 0 k
Fik dV0
x
J
x
J
w
w
i
i
V0
&
It was proven in Chapter 2 that x& J 1 F 0 , thus, the above equation becomes:
(5.121)
&
x
&
qdV
V0
wq 0 k 1
Fik dV
wx i J
V0
wq 0 k wx i
wxi wX k
dV0
V0
wq 0 k
wX k
dV0
(5.122)
&
X& q 0 dV0
V0
Bearing in mind the equations in (5.116), (5.117), (5.118) and (5.122), the energy equation
in the reference configuration can be established as:
V0
u dV0
S : E
&
X
q 0 S 0 r dV0
&
(5.123)
V0
&
&
S 0 u ( X , t ) S : E X& q 0 S 0 r ( X , t )
5.9.7
(5.124)
In this subsection we obtain the energy equation for a domain with a singular surface 6(t )
as discussed in subsection 5.5.1, (see Figure 5.4). In this case the energy equation becomes:
DK DU DW DQ
Dt
Dt
Dt
Dt
(5.125)
& &
&
& &
&
D 1
S (v v )dV SudV
Sb v dV v ndS S rdV q n dS
Dt 2 V 6
V 6
V 6
V 6
S S
S S
317
For the terms on the left of the equation in (5.125) we can apply Reynolds transport
& &
theorem, (see the equation in (5.22)), to ) S (v v ) Su , thus:
1
& &
S (v v ) Su dV
2 V 6
1 & &
D S 2 (v v ) Su
& &
&
& &
& &
S 12 (v v ) Su x& v dV >>S 12 (v v ) Su v Z@@ n dS
Dt
V 6
6
D
Dt
(5.126)
Then by mathematically manipulating the terms of the volume integral we can see that:
& &
D S 12 (v v ) Su
& &
&
S 12 (v v ) Su x& v
Dt
& &
&
DS
Du
u
S
S 12 (v v ) Su x& v
Dt
Dt
1
2
& &
D 1 (v v )
& & DS
(v v )
S 2
Dt
Dt
(5.127)
& &
&
S (v v ) Su x& v
1
2
& &
D 12 (v v )
Dt
Dt
&
DS 1 & &
Du 1 & &
(v v ) u S 2 (v v ) u S x& v
Dt
Dt 2
& &
D 1 (v v )
12 (v& v& ) u DS S x& v& S Du S 2
Dt
Dt
Dt
(5.128)
Thus,
& &
D S 12 (v v ) Su
Dt
& &
&
S 12 (v v ) Su x& v
& &
Du
Sv v
Dt
(5.129)
Then if we return to the equation in (5.126), and if we refer to (5.129) we can conclude
that:
D
Dt
1
& &
S (v v ) Su dV
2 V 6
& &
& &
Du & &
S
v v dV >>S 12 (v v ) Su v Z@@ n dS
Dt
V 6
6
(5.130)
For the surface integrals on the right side of the equation in (5.125) we can apply Gauss
theorem to a volume with discontinuity, (see equation (5.17)):
&
v q n dS
&
S S
&
&
&
&
v q dV >>v q@@ n dS
&
x
V V
V ij , j vi V ij v i , j x&
&
(5.131)
&
&
v x& : x& v , and if
we have observed the spatial velocity gradient has been broken down into a symmetric and
&
an antisymmetric part, we obtain : x& v : D : W : D , so, we can conclude that:
&
v q n dS
&
S S
v
&
V V
&
x
&
&
v& x& : D x& q dV
V V
&
&
>>v& q& @@ n dS
(5.132)
318
Then by substituting the equations (5.130) and (5.132) into the energy expression in (5.125)
we obtain:
& &
& &
Du & &
v v dV >>S 12 (v v ) Su v Z@@ n dS
Dt
6
&
&
&
v& x& : D x& q dV >>v& q@@ n dS Sb v&dV S rdV
V 6
V 6
(5.133)
or
& &
&
&
Du & &
v v v x& : D x& q Sb v S r dV
Dt
V 6
&
& &
& &
&
>>S (v v ) Su v Z v q@@ n dS
1
2
V 6
(5.134)
&
&
&
&
Du
Sv x& Sb dV
: D x& q S r v
&
Dt
&
& &
& & &
>>S 12 (v v ) Su v Z v q@@ n dS
(5.135)
Su : D
&
x
V 6
& &
&
&
&
&
(5.136)
which thereby results in the energy equation for volumes with discontinuity:
&
Su : D x& q S r
in V
>>S 12 (v& v&) Su v& Z& v& q& @@ n 0
on 6
(5.137)
319
S I ( x, t ) dV S
H (t )
&
I ( X , t ) dV0
V0
J
K
(5.138)
J
sK
(5.139)
J
.
kgK
&
Sb ( x, t )dV S
&
b ( X , t ) dV0
V0
where b is the source of local entropy per unit mass per unit time >b @
J
.
kg s K
Then the entropy flux that enters the system through the material surface is defined by:
&
s n dS
J
sK
S s&
(5.140)
&
&
D
&
S I ( x , t ) dV t Sb ( x , t )dV s n dS
Dt V
V
S&
&
&
S I ( x , t ) dV t Sb ( x , t )dV
(5.141)
&
s n dS
S s&
S I dV t SbdV
&
x
s& dV
(5.142)
NOTE: The global form of the entropy inequality in (5.142) implies that: if entropy occurs
then the process is irreversible, that is, we can not return to the original system without
adding work to the system. And, the equality of (5.142) represents a reversible process.
The local form of the equation in (5.142) is given by:
&
&
SI ( x, t ) t Sb x& s
(5.143)
&
q & (1)
s
T
r
b (1)
T
&
(5.144)
&
where T ( x , t ) t 0 is the absolute temperature, >T @ K , and by assuming that s (1) and b (1)
are equal to zero, the entropy inequality in (5.143) becomes:
SI t S
Thus,
r
x&
T
&
&
r 1
1 &
q
S x& q 2 q x& T
T T
T
T
(5.145)
320
&
&
r ( x, t )
q
x& t 0
T
T
&
&
r ( x, t ) 1 & &
1 &
SI ( x , t ) S
x q 2 q x& T t 0
T
T
T
&
SI ( x , t ) S
Entropy inequality
(current configuration)
(5.146)
We can also express the entropy inequality given in (5.143) in the reference configuration
as:
&
& &
&
S 0 I ( X , t ) t S 0b ( X , t ) X& S ( X , t )
&
(5.147)
where S is the entropy flux vector in Lagrangian description. For thermal processes, the
entropy flux vector and entropy source can be established, respectively, as:
& &
S ( X , t)
&
&
q0
S1
T
&
b ( X , t)
&
r( X , t)
b1
T
&
(5.148)
Then if we take into account the equation in (5.148) where S 1 and b1 are equal to zero,
the equation in (5.147) becomes:
&
&
&
q
r( X , t)
S 0 I ( X , t ) t S 0
X& 0
T
T
&
&
r( X , t) 1 & &
1 &
S 0 I ( X , t ) t S 0
X q 0 2 q 0 X& T
T
T
T
Entropy inequality
(reference configuration)
&
&
J q k Fik1 , it is true that:
&
& wT
q 0 X& T q 0 i
wX
i
Material
(5.149)
&
J q F T , or in indicial notation
&
&
&
& wT
wT
wT wx p
wT
J qk
F pi J q k E pk
J q k Fik1
J q k Fik1
wx k
wx p
wx p
wx p wX i
Spatial
& &
&
J q( x , t ) x& T ( x , t )
(5.150)
SI S
(5.151)
In this scenario, the entropy inequality is called the Clausius-Duhem inequality, and is given by:
&
S I ( x , t )
1
1
1 &
: D S u 2 q x& T t 0
T
T
T
(5.152)
321
We can also express the Clausius-Duhem inequality in the reference configuration. From
&
&
the equation in (5.124) we obtained S 0 r ( X , t ) X& q 0 S 0 u S : E and by substituting
this into the entropy inequality given in (5.149) we obtain:
&
&
&
1
1 &
S 0 r ( X , t ) X& q 0 2 q 0 X& T
T
T
&
1
1 &
S 0 I ( X , t ) t S 0 u S : E 2 q 0 X& T
T
T
S 0 I ( X , t ) t
(5.153)
or:
&
S 0 I ( X , t )
1
1
1 &
S : E S 0 u 2 q 0 X& T t 0
T
T
T
or
(reference configuration)
&
1
1
1 &
S 0 I ( X , t ) P : F S 0 u 2 q 0 X& T t 0
T
T
T
(5.154)
Note that the inequality q x& T d 0 is always valid, since the orientation of the heat flux
&
vector ( q ) is always opposite to the temperature gradient ( x& T ), (see Figure 5.10). Then,
we can formulate the heat conduction inequality:
Heat conduction inequality
&
q x& T t 0
(current configuration)
&
q 0 X& T t 0
(reference configuration)
(5.155)
&
Clausius-Planck
inequality
1
T
&
(5.156)
configuration)
where Dint is the internal energy dissipation, which requires positiveness at any time,
D int t 0 .
Z u TI
J
kg
(5.157)
322
&
&
&
qn (q n )n qn n
x& T x& T n
&
&
q x& T x& T q n
x& T
s
T1
&
qn
T1 ! T2 ! T3
T2
T3
&
q x& T
& &
&
(qn q s ) x& T qn x& T
u IT Z
TI
TSI
>
(5.158)
Then if we consider that T ! 0 (absolute temperature) and the entropy inequality given in
(5.145), we obtain:
SI t S
r 1 & &
1 &
x q 2 q x& T
T T
T
&
STI t S r x& q
1&
q x& T
T
(5.159)
Afterwards by combining the above inequality with the equation in (5.158) we obtain:
>
& 1&
Su S IT Z t S r x& q q x& T
T
(5.160)
&
>
&
t S r & q& 1 q& & T
: D x& q S r S IT Z
x
x
T
t 1 q& & T
: D S IT Z
x
T
>
(5.161)
>
1 q& & T t 0
: D S I7 Z
x
T
Clausius-Duhem inequality
(current configuration)
(5.162)
323
S 0TI S : E S 0 u
1&
q 0 X& T t 0
T
1&
S : E S 0 >u TI @ q 0 X& T t 0
T
(5.163)
can also be written in terms of the Helmholtz free energy Z . To do this let us consider the
&
&
&
Helmholtz free energy in Lagrangian description Z u ( X , t ) T ( X , t )I ( X , t ) . Additionally,
the rate of change is given by Z u TI TI u TI Z TI with that the ClausiusDuhem inequality in the reference configuration becomes:
>
>
@
@
D (S 0 Z )
Dt
D(S 0 )
D (Z )
S0
Dt
Dt
S 0 Z .
Form
of
the
Clausius-Planck
The Clausius-Planck inequality can also be expressed in terms of Helmholtz free energy.
&
Then, if we consider the heat conduction inequality, q x& T t 0 , the equation in (5.162)
becomes:
>
(5.165)
(5.166)
Dint : D S IT Z t 0
>
Dint S : E S 0 IT Z t 0
The work done by the forces between two points being independent of the path;
If we take into account that the dissipation of energy is equal to zero in a reversible process
we obtain:
Dint : D SZ 0
SZ : D
(5.167)
Dint S : E S 0 Z 0
1
S : C S 0 Z
2
S 0 Z
1
S : C
2
S : E
(5.168)
324
&
&
D
&
S I ( x , t ) dV t
Sb ( x , t )dV s n dS
Dt V V
V V
S S
(5.169)
For the surface integral on the right side of the inequality in (5.169) we can apply the
divergence theorem with discontinuity given in Eq. (5.17), the result of which is:
&
s n dS
S S
&
&
s dV >>s @@ n dS
V V
(5.170)
For the volume integral on the left side of the inequality in (5.169) we can apply the
Reynolds transport theorem given in (5.22) in which ) S I , then:
D
Dt
S I dV
V 6
&
& &
DS I
S I x& v dV >>S I v Z@@ n dS
Dt
V 6
6
(5.171)
Then by substituting the equations in (5.171) and (5.170) into (5.169), we obtain:
&
& &
&
DS I
Sb ( x , t )dV
S I v dV >>S I v Z @@ n dS t
Dt
V 6
6
V V
&
&
x& s dV >>s @@ n dS
V 6
(5.172)
I
S I x& v x& s Sb dV >>S I v Z s @@ n dS t 0
S
Dt
Dt
V 6
6
(5.173)
Note that the equation I
&
DS
S I x& v
Dt
&
DS
S x& v 0 is valid due to the mass
Dt
x& s Sb dV >>S I v Z s @@ n dS t 0
S
Dt
V 6
6
(5.174)
(5.175)
Problem 5.11: 1) Consider a continuum motion in which the stress power is equal to zero.
&
Also, consider that the heat flux is given by q K (T ) x& T , which is known as Fouriers
law of thermal conduction, where K (T ) is a second-order tensor called the thermal conductivity
tensor (the thermal property of the material), and c
wu (T )
, where c is the specific heat
wT
capacity at a constant deformation (the thermal property of the material) and is expressed in
325
J
. Taking into account all previous considerations, find
K
the energy equation for this process. Then also provide the unit of K (T ) in the
: D x& q Sr x& q Sr
,
wT wt
0
&
wT
Sc
x& > K (T ) x& T @ Sr
x& q Sr
wt
S u S
Sc
or
wT
wt
Sc
wT
wt
The above equation is called the heat flux equation which is applied to the thermal
conduction problem.
Then if we take into account the following units: >q@
&
J
m2s
W &
wT
, xT { &
wx
m 2
K
, we
m
can ensure that the units are consistent if the following is met:
>q& @
J
m2s
W
m 2
W K
J
s m K m K m
W
J
s m K m K .
NOTE: As we will see later, when the stress power is equal to zero, we can decouple the
thermal and mechanical problem. That is, we can study these problems separately.
2) a) We start from the heat conductivity inequality:
&
&
q x& T (K ( x ) x& T ) x& T t 0
&
x& T K ( x ) x& T t 0
or
q i T,i
( K ij T, j )T,i t 0
T,i K ij T, j t 0
&
&
>
x& T x& T K
skew
&
xT
t0
x& T t 0
326
Notice that K skew : ( x& T x& T ) 0 , since the double scalar product between an
antisymmetric tensor ( K skew ) and a symmetric one ( x& T x& T ) is equal to zero, then:
&
0 d x& T K ( x ) x& T
b) For the proposed problem the only remaining governing equation is the energy
&
&
Du
equation: S
{ Su : D x& q Sr x& q , where u is the specific internal
Dt
energy, : D is the stress power, and Sr is the internal heat source per unit volume. Then:
Su q i ,i (K ij T, j ) ,i K ij ,i T, j K ij T, ji ( x& K T ) ( x& T ) K : x& ( x& T )
>
sym
T, ji . If the material is
&
homogeneous the implication is that the K field does not depend on ( x ) , so K ij ,i 0 j . In
T,ij
&
DS
S ( x& v ) 0 (1 equation)
Dt
&
&
x& Sb Sv (3 equations)
T
(6 unknowns)
&
S u : D x& q Sr (1 equation)
&
SI ( x , t )
1
1
1 &
: D S u 2 q x& T t 0
T
T
T
(5.176)
(5.177)
(5.178)
(5.179)
(5.180)
327
The entropy inequality is not one more problem equation. Rather, it is used to establish
restrictions on the problem variables. The symmetry of the Cauchy stress tensor reduces
the number of -unknown from 9 to 6.
The mass continuity equation, the equations of motion and the energy equation give us in
&
total 5 equations. The unknowns are: the three components of velocity v , temperature T ,
mass density S , six components of the Cauchy stress tensor , the specific internal energy
&
u , three components of the heat flux vector q , and the entropy I , with a total of 16
unknowns.
To achieve the well-posedness of the problem eleven equations must be added. We must
add equations that connect the stress, heat and energy with other fields. These equations
are called constitutive equations, which is the subject of the next chapter.
&
&
&
For problems which only involve the transport of a physical quantity (mass, energy, or
otherwise) the only principle necessary to establish the governing equations is the
conservation law of the physical quantity, or in its strong form: the physical quantity
continuity equation, (see equation (5.24)):
Q
&
&
& w) ( x, t )
w) ( x , t )
x&
x& ()v ) {
wt
wt
& &
q( x, t )
>) @
s
(5.181)
The case in Problem 5.11 was related to energy transport (not mass transport). Said energy
transport exists due to the agitation of atoms, in which the degree of agitation at the
macroscopic level is characterized by the temperature. If a particle starts to increase the
degree of agitation, then neighboring particles also start behaving in a similar fashion. In
this way the energy in solids is transported without any mass transport. This energy
&
&
transport at the macroscopic level is represented by means of the flux, q )v .
When we are working in the field of continuum mechanics we do not go down to the
atomic level and measure the average velocity (vibration) of a handful of atoms to establish
the flux. What we do is: we go to the laboratory with the material with which we want to
establish the heat flux (energy flow), we vary the temperature and we verify
macroscopically that the flux can be characterized by the following phenomenological law
q kT (in a one-dimensional case), where k is a thermal property of the material. This
procedure was performed by Fourier, thereby establishing Fouriers law of heat conduction.
Fourier also verified in the laboratory that heat flux is opposite to the temperature gradient,
a fact already proven by the second law of thermodynamics. The law q kT is a
phenomenological law or constitutive equation of heat flux, and connects two thermal
328
kg
J
, >c @
with which we can verify the following SI
kg K
m3
units:
>) @ >ScT @
>q& @ { >)v& @
kg J
K
m 3 kg K
J m
m3 s
J
(unit of energy per unit volume - energy density)
m3
J
(unit of energy flux)
m2s
There are several engineering problems which are characterized by the continuity equation,
some of which are: heat conduction problems (energy flux); filtration problems in porous
media (mass transport); diffusion problems (e.g. transport of contaminant in an aqueous
medium); and the Saint-Venant torsion problem (stress flux).
K x& T
J
m2s
(5.182)
where q is the heat flux per unit area per unit time, and its SI unit >q@
&
&
329
J
m2s
W
; x& T is
m2
K
, and K is the thermal conductivity
m
W
, (see Problem 5.11).
mK
NOTE: Fouriers law of heat conduction is not universal as there are complex materials in
which heat flow is governed by more complex laws.
The negative sign in Fouriers law is there because the heat flux vector is always opposite to
the temperature gradient. The temperature gradient vector ( x& T ) points from the coldest
&
to the warmest region, while the heat flux vector ( q ) points from the warmest to the
coldest region (physical fact), (see Figure 5.10).
conduction
conduction
K 11 K 12
K
21 K 22
K 31 K 32
K 13
K 23
K 33
isotropic material
o
(K ) ij
1 0 0
K 0 1 0
0 0 1
(5.183)
For isotropic materials, i.e. those that have the same property in any direction, the thermal
conductivity tensor is represented by a spherical tensor, (see Problem 5.11).
&
&
(q) i
wT
wx1
q1
1 0 0
wT
q
K
2
0 1 0 wx
q
0 0 1 2
3
wT
wx3
wT
wx1
wT
K
wx 2
wT
wx3
(5.184)
330
q i n i
&
q n , (see Figure 5.2), is evaluated as:
q1n 1 q 2 n 2 q 3n 3
K
wT
wT
wT
n1 K
n2 K
n3
wx1
wx 2
wx3
(5.185)
BT Text
(5.186)
where q is the thermal energy; B is the heat transfer coefficient per unit area; T is the
temperature of the bodys surface, and Text is the temperature of the surrounding
environment.
warm air
radiator
cold air
Heat
generated
internally
Heat that
leaves the
system
Change of
internal
energy
331
Let us consider a differential volume element, (see Figure 5.13), in which there are inflows
and outflows of heat. In addition let us consider energy generated internally represented by
Q Sr (per unit volume per unit time), whose SI unit is >Q @
J
. The scalar function r
m3s
describes the heat generated which could be caused by a phenomenon such as a chemical,
or nuclear reaction, and whose SI unit is >r @
J
. As the temperature of the body
kg s
increases, part of the thermal energy is stored in the body. For a differential volume
element ( dx1 dx 2 dx3 ) this stored energy is governed by the expression:
Sc v
wT
dx1 dx 2 dx3
wt
(5.187)
where S is the mass density; and the material property c v is the specific heat capacity at a
constant volume whose SI unit is >c v @
J
.
kg K
qz
wq z
dz
wz
qy
wq y
wy
dy
dy
qx
qx
Q
wq x
dx
wx
dz
y
x
qy
dx
qz
q x dy dz q y dx dz q z dx dy Q dx dy dz q x x dx dydz
wx
wq y
wq
wT
q y
Sdxdydz
dy dxdz q z z dz dxdy c v
w
w
wt
y
z
wq y wq z
wq
wT
Q x
Sc v
w
w
x
y
w
z
wt
(5.188)
K x& T Scv
wT
wt
(5.189)
332
K ij
wT
. Notice that the
wx j
above equation was obtained in Problem 5.11 and it should be pointed out that we have
one equation in (5.189) and one unknown (temperature). The solution of equation (5.189)
is unique if we are given the appropriate boundary and initial conditions. The governing
equation in (5.189) together with the boundary and initial conditions are called the Initial
Boundary Value Problem (IBVP) of thermal conduction.
Then by considering an isotropic homogeneous material, the heat equation in (5.189)
becomes:
Q K
w wT
w wT
K
wx wx
wy wy
w wT
wT
K
S cv
w
z
w
z
wt
Q w 2 T w 2 T w 2T
K wx 2 wy 2 wz 2
(5.190)
1 wT
(5.191)
L wt
K
S cv
Particular Cases
(5.192)
&
(5.193)
Q
2x& T
K
(5.194)
From a mathematical point of view, the above equation is known as the Poissons
equation.
(5.195)
Q 0
w 2 T w 2T w 2 T
0 2x& T 0 Laplaces equation
(5.196)
wx 2 wy 2 wz 2
&
Transient problem, T T ( x , t ) (time dependent), but in the absence of internal heat
generation, Q 0 , the equation in (5.191) becomes Fouriers equation:
w 2 T w 2 T w 2T
wx 2 wy 2 wz 2
1 wT
L wt
2x& T
1 wT
L wt
Fouriers equation
(5.197)
T ( x, y , z , t ) T *
t !0
to
333
(5.198)
on S1
wT
wT
wT
nx K
ny K
nz
wx
wy
wz
to
t !0
on S 2
(5.199)
wT
wT
wT
nx K
ny K
n z B (T Text ) 0
wx
wy
wz
t!0
to
on S 3
(5.200)
where n x , n y , n z are the components of the outward unit normal vector on the surface.
Initial conditions T ( x, y, z , t 0) T0 .
&
q*
t
S3
x3
dV
Sr
x2
J E
D
B
Text
T*
S1
x1
n x
n y
n z
cos D
cos E
cos J
334
b
dam (impermeable)
h1
material point - P
h2
Flux
qy
wq y
wy
dy
qx
qx
dy
wq x
dx
wx
dx
soil (permeable)
qy
x
rock (impermeable)
wq y
wq y
wq
wq x
dx dy q y q y
dy dx 0 x dxdy
dxdy
q x q x
w
wy
w
w
x
y
x
(5.201)
The phenomenological law of mass flux in porous media is governed by Darcys law,
&
q K x& G , where G is the total potential (water level), and K is the permeability tensor
&
which depends on the material. In the two-dimensional case, the components of q are
given by:
qx
K
wG
wx
qy
K
wG
wy
(5.202)
where we have considered an isotropic material (one which has the same permeability in all
directions). Then by substituting the components of the flux vector in Equation (5.201), we
obtain:
w wG w wG
0
K
K
wx wx wy wy
w 2 G w 2G
K 2 2 0 2 G 0 Laplaces equation
wy
wx
(5.203)
(5.204)
335
wG
wx
wG
wy
wG
( x, L ) 0
wy
wG
wx
x f
0
x f
0
y 0
b
b
dxd
2
2
x
h1
b
2
G ( x, L )
x!
b
2
h2
(5.205)
mol
m2s
(5.206)
&
where D ! 0 is the diffusion tensor (or diffusivity tensor), and c( x , t ) is the solute
concentration whose SI unit is >c @
c
mol
. This concentration is defined as follows:
m3
solute mass
fluid mass
(5.207)
1 solute mass
volume
Sf
Ss
Sf
(5.208)
where S f , S s are the mass densities of the fluid and solute, respectively.
In general, when we have a process where there is mass transport (a fluid+solute) two
mechanisms take place, namely: convection and diffusion. In this case the matter (solute) is
defined by the concentration c and we must consider the matter to be diffused throughout
the aqueous medium. Here, we can assume that the amount of the matter is too little to
affect the fluid velocity field.
&
&
Let us consider the solute flux q c v (a convective term) to which we add the diffusive
term to obtain the total flux:
&
q
wc
D &
w
x
Convective
&
c,v
term
(5.209)
Diffusive
term
Then, to obtain the partial differential equation for the convection-diffusion problem we
consider the one-dimensional case, (see Figure 5.16).
336
qx
qx
wq x
dx
wx
x dx
solute
generated
internally
solute that
leaves the
system
Change of
the solute
internally
q x dy Qdxdy q x x dx dy
wx
wc
dxdy
wt
Q
wq x
wx
wc
wt
(5.210)
Additionally, by substituting the flux given in (5.209) into the above equation, we obtain:
wc
w c v x D
wx
Q
wx
wc
wt
Q
wc v x w wc
D
wx
wx wx
wc
wt
(5.211)
wc
wt
Convection-diffusion
equation
(5.212)
&
wc
is the local variation of the concentration with respect to time, v x& c is the
wt
convection term caused by fluid motion, and x& (D x& c) is the diffusion.
where
Next, we assume that at a material point there are two types of material that are
represented by a physical quantity per unit volume in such a way that c c f c s , and the
& &
&
following holds v v f v s , (see Figure 5.17).
cf
&
vf
material point - P
V
cs
&
v
&
vs
337
Then, from the continuity equation for this physical quantity we obtain:
Q
w)
x&
wt
&
)v
>
&
&
w (c f c s ) w
& (c f c s )(v f v s )
wx
wt
(5.213)
thus
>
&
&
w (c f c s ) w
& (c f c s )(v f v s )
wt
wx
&
&
&
w (c f c s ) w f & f
Q
& c v c f v s csv f csv s
wt
wx
&
&
&
&
wc f wc s
Q
x& c f v f c f v s c s v f c s v s
wt
wt
wc f
& wc s
&
&
&
Q
x& (c f v f )
x& (c s v f ) x& c f v s c s v s
w
t
w
t
>
>
(5.214)
>
&
wc f
x& (c f v f ) 0 holds,
wt
which is the continuity equation of the quantity c f with which the equation in (5.214)
becomes:
>
&
&
&
wc s
x& (c s v f ) x& c f v s c s v s
wt
&
&
&
wc s
(5.215)
x& (c s v f ) x& (c s v s ) x& (c f v s )
Q
wt
&
&
&
&
wc s
Q
x& (c s v f ) x& (c s v s ) x& c f v s c f x& v s
wt
&
If the physical quantity c f does not change with x , then the gradient of c f becomes
&
x& c f 0 . In addition if we consider the medium ( s ) to be incompressible we obtain
&
x& v s 0 . These simplifications indicate that the material ( s ) does not affect the velocity
field of the material ( f ). So, if the amount of the material ( s ) is significant, this approach
Q
(5.212).
338
D x& G S c
wG
wt
(5.217)
Heat flux
Diffusion
Saint-Venant
torsion
Scalar field
G
Temperature
T
piezometric head
(or hydraulic head)
h
ion concentration
c
Prandtl stress
function
D x& G
Thermal
conductivity
tensor
Heat
generated
Flux vector
&
q
Heat flux
vector
Phenomenological
law
Fouriers law
&
q
&
q K x& T
permeability
tensor
water
source
Volume flux
vector
&
q K x& G
Constitutive
matrix for
diffusion
coefficient
Ion source
1
G
2T
Ion flux
vector
&
J
Darcys law
Ficks law
&
J D x& c
Hookes law
339
It must be stressed that until the fundamental equations were established, we did not
discuss the type of material that the continuum is made up of. When we begin to specify
this is the time when the concept of the constitutive equation appears. Then, we can
automatically believe that every class of material has its particular constitutive equations.
In order to represent the real behavior of the material, the constitutive equation has to be
calibrated with macroscopic parameters, i.e. these parameters, which are obtained in the
laboratory, represent the material behavior at the macroscopic level. Remember that the scale
of study of continuum mechanics is macroscopic, and then we need to obtain some
representative macroscopic parameters of the phenomena that occur at the microscopic
scale. We can consider this to be the Achilles heel of Continuum Mechanics, because in
some cases we are not able to obtain a macroscopic parameter that characterize
phenomena that are taking place at the microscopic level. In fact, the evolution of the
constitutive equation is directly linked to the precision of the instrumentation and new
techniques used in laboratory testing of such materials. So in summary we can state that the
constitutive equations used to characterize a material must be capable of simulating any
phenomena that arise in the material during the loading/unloading/loading process, or at
least the most significant.
As discussed earlier, the constitutive equations complete the set of equations that govern a
particular physical problem. That is, they complement the IBVP. The solution of the
problem can be analytical (the exact solution) or numerical (an approximate solution). In
most cases it is impossible to obtain the analytical solution, so we turn to computational
mechanics to obtain the numerical solution of the physical problem.
Computational Mechanics resolve specific problems by using numerical simulation tools
incorporated in the computer. In general we can state that computational mechanics is not
an independent block, i.e., for its complete implementation it is directly dependent on three
areas: Theoretical Analysis (IBVP establishment), Experimental Analysis (Lab), and Numerical
Analysis (a numerical methodology incorporated into the computer to obtain the numerical
solution for IBVP), (see Figure 5.18). From a very general point of view, we can also
appreciate in Figure 5.18 how the constitutive models are embedded within the field of
Computational Mechanics.
340
COMPUTATIONAL MECHANICS
STRUCTURE
Proposition for a
CONSTITUTIVE
MODEL
LABORATORY
Testing proposal
If possible
NUMERICAL SOLUTION
Input data
Option 3:
New numerical
method
Option 1
NO
Option 2
NO
YES
Is the simulation
realistic?
Option 4:
New IBVP proposal.
The Continuum Theory
is not suitable.
Numerical Simulation
New testing
proposal
Introduction to
Constitutive Equations
6.1 Introduction
force/moment
Constitutive law
V
strain
displacement
341
342
&
DS
S ( x& v ) 0
Dt
(6.1)
&
&
x& Sb Sv
(6.2)
(6.3)
&
S u : D x& q Sr
&
SI ( x , t )
1
1
1 &
: D S u 2 q x& T t 0
T
T
T
(6.4)
(6.5)
D
( JS ) 0
Dt
&
X& P S 0 b 0
X&
(6.6)
&
S 0V
&
&
F S S 0 b 0 S 0V
S S T or P F T F P T
&
&
&
S 0 u ( X , t ) S : E X& q 0 S 0 r ( X , t )
&
&
&
or S 0 u ( X , t ) P : F X& q 0 S 0 r ( X , t )
1
1
1 &
S : E S 0 u 2 q 0 X& T t 0
T
T
T
1
1
1 &
or S 0 I P : F S 0 u 2 q 0 X& T t 0
T
T
T
(6.7)
(6.8)
(6.9)
S 0 I
(6.10)
As we saw in Chapter 5, the mass continuity equation, the equations of motion and the
energy equation give us in total 5 equations. The unknowns are: three components of
&
velocity v , temperature T , mass density S , six components of the Cauchy stress tensor
&
T , the specific internal energy u , three components of the heat flux vector q , and
343
(6.11)
(6.12)
(6.13)
where the constitutive equations for stress provide six equations, the heat conduction law
provides three equations, and thermodynamic state laws provide two equations, which
results in a well-posed system, with 16 unknowns and 16 equations.
The equations that relate state functions to state variables are called the equations of state or
constitutive equations and state variables, the selection of which depends on the problem in
hand, are those that depend only on themselves. For instance, when we are dealing with
solids, in general, we use strain and temperature as independent variables. In this scenario,
in thermodynamics, internal energy u (or Helmholtz free energy Z ), entropy I , heat flux
&
q , and the Cauchy stress tensor are all considered to be state functions, which can be
determined by the state variables. Then, how a material responds is fully defined by the
&
fields ( Z , , I and q ). Depending on the problem it may be more appropriate to use
other thermodynamic potentials among: u ( E , I ) -specific internal energy, Z ( E , T ) -Helmholtz free
energy, H(S, T ) -enthalpy or G(S, T ) -Gibbs-free energy. For further details concerning these
potentials see Chapter 10. On a final note, the effects caused by an electromagnetic or
chemical change will not be considered here.
NOTE: It must be emphasized that as constitutive equations describe material
constitutions of systems from a macroscopic point of view, based on experimental
evidence, then, constitutive equations are, by their nature, approximations.
344
6.2.1
&
This principle states that the fields ( Z , , I , q ) at a material point ( X ) depend on the
&
& &
entire history of the motion, x ( X , t ) , and the temperature history, T ( X , t ) , i.e., up until the
&
present time t , but they never depend on the future values of ( x , T ).
In certain types of thermodynamic processes, it may be unrealistic for the current field
&
values ( Z , , I , q ) to depend on values that are too distant from the current ones,
hence, we have the principle of limited memory. The history of fields that are too far removed
from the current ones do not affect them. So, we must only consider the latest values of
the fields, which implies we need to define what is meant by recent time.
6.2.2
This principle states that the current values of the fields ( Z , , I , q ) at a material point
depend on the state of these field in the vicinity of the point. Motion information is given
&
locally by the deformation gradient F ( X , t ) , and temperature by its gradient T .
Then, the materials that satisfy the determinism and local action principles are called simple
thermoelastic materials.
6.2.3
This principle states that: there is no reason to exclude a priori a state variable (independent
variable) of the constitutive equations. For example, if we have a simple thermoelastic
material it makes no sense if stress is only a function of the deformation gradient while the
heat flux vector is only a function of temperature.
6.2.4
This principle states that the constitutive equations must be the same for any observer.
Therefore, any constitutive equation must satisfy the principle of objectivity, (see Chapter
4). That is, if an observer detects a stress state in the body B which undergoes a rigid body
motion, then, this observer must detect the same stress state in the body B * , (see Figure
6.2).
6.2.5
This principle states that: constitutive equations must satisfy the entropy inequality for any
admissible thermodynamic process.
&
x*
observer
&
&
c (t ) Q(t ) X
345
B*
*
Z(t ) Z ( X , F ( W) , T ( W) , X& T ( W) )
&
I (t ) I ( X , F ( W) , T ( W) , X& T ( W) )
&
& &
q 0 (t ) q 0 ( X , F ( W) , T ( W) , X& T ( W) )
&
P (t ) P ( X , F ( W) , T ( W) , X& T ( W) )
(6.14)
where x (W) represents the history of x , until the present time t , W d t . We can also verify
&
that Z and I are scalar-valued functionals, q is a vector-valued functional and P is a
second-order-valued functional. Then, taking into account the principle of dissipation, the
Clausius-Duhem inequality must be satisfied for any thermodynamic process.
For a homogeneous simple thermoelastic material, the functionals described in (6.14) are
&
independent of X :
Z(t ) Z ( F ( W ) , T ( W ) , X& T ( W ) )
I (t ) I ( F ( W) , T ( W) , X& T ( W) )
&
&
q 0 (t ) q 0 ( F ( W) , T ( W) , X& T ( W) )
(6.15)
P (t ) P ( F ( W) , T ( W) , X& T ( W) )
NOTE: The functions with hat x (functional) are distinct from those that are on the left
of the equation, i.e., x provides the current value of x (t ) taking into account the entire
history of the arguments of x .
346
According to the principle of objectivity, the constitutive equations must be invariant under
rigid body motion. Then the constitutive equations must satisfy the following:
*
*
Z(t ) Z ( F ( W) , T ( W) , X& T ( W) )
*
I (t ) I ( F ( W) , T ( W) , X& T ( W) )
&
&
*
*
q*0 (t ) q 0 ( F ( W) , T ( W) , X& T ( W) )
(6.16)
*
*
P * (t ) P ( F ( W) , T ( W) , X& T ( W) )
wT
(6.17)
w X& T
>
(6.18)
where Z is the Helmholtz free energy per unit mass, K is the entropy per unit mass, and
P is the first Piola-Kirchhoff stress tensor. Note that, although the stress power P : F is
defined in the reference configuration, neither P nor F are in the reference configuration.
Then, by combining the equation in (6.17) with the entropy inequality in (6.18) we obtain:
.
&
wZ wZ wZ
P : F S 0
T TI 1 q 0 X& T t 0
:F
T
wT
wT
wF
T
.
&
wZ
wZ
wZ
: F S0
T 1 q 0 X& T t 0
I T S 0
P S 0
wT
T
wF
wT
(6.19)
The above inequality must be satisfied for any admissible thermodynamic process.
Let us now consider the process such that F 0 , and a system with uniform temperature,
&
&
0 . In this particular thermodynamic process the inequality in (6.19)
wZ
I T t 0
S0
wT
(6.20)
Note that the inequality in (6.20) must also be met for any thermodynamic process. Then,
if in the current process the condition in (6.20) is met, we can then apply another process
such that T o T , in which the entropy inequality is violated. Thus, the only way in which
the inequality in (6.20) is satisfied is when:
wZ
I
wT
I
wZ
wT
(6.21)
. 1&
wZ
wZ
&
P S 0 wF : F S 0 wT T T q 0 X T t 0
347
(6.22)
Now let us consider a process where F 0 , with which the inequality in (6.22) becomes:
S0
.
&
wZ
T 1 q 0 X& T t 0
wT
T
(6.23)
&
1 &
q 0 X& T t 0 is always true, since the heat flux vector ( q 0 ) is always
T
wZ
z 0 we have an inconsistency, i.e.
opposite to the temperature gradient ( X& T ). If
w X& T
we can use X& T in such a way that the condition in (6.23) is violated, with which we can
conclude that Z should not depend on the temperature gradient, i.e. Z Z ( F , T ) . Then
the entropy inequality in (6.22) becomes:
wZ 1 &
&
(6.24)
P S 0 wF : F T q 0 X T t 0
&
Now let us consider a process where X& T 0 (a uniform temperature field), then the
P S 0 wF : F t 0
(6.25)
Starting from this point, we could apply another process where F o F , thus:
wZ
P S 0
:F t0
wF
(6.26)
Then, the only way that the two equations (6.25) and (6.26) can remain valid is when:
P S0
wZ
wF
P S0
wZ
wF
(6.27)
Afterwards, we can conclude that the constitutive equations for a simple thermoelastic
material are given by:
Z Z(F , T )
wZ ( F , T )
Constitutive equations for a simple
wF
thermoelastic material
wZ ( F , T )
I(F , T )
wT
&
&
q 0 q 0 ( F , T , X& T )
P(F , T ) S 0
(6.28)
As we state before, the constitutive equations must satisfy the principle of objectivity and
any scalar, for instance energy and entropy, satisfies this principle. However, we can take
advantage of this principle in order to express these scalars in terms of other appropriate
parameters. Then by applying the principle of objectivity to the energy we have:
Z * Z(F * , T * )
Z (Q F , T )
(6.29)
348
where we have taken into account that F * Q F and T * T (see Chapter 4). Then as the
principle of objectivity must be met for any orthogonal tensor Q , we use the transpose
rotation tensor ( Q R T ) of the polar decomposition ( F R U ) as the orthogonal tensor,
(see Figure 6.3), with which we obtain:
Z Z (Q F , T )
Z (R T R U, T )
Z (U, T )
(6.30)
That is, to satisfy the principle of objectivity, the energy must be a function of the right
stretch tensor. Then, if we take into account the equations C U 2 and C 2 E 1 , (see
Chapter 2), we can still express the energy in terms of C or E , i.e.:
Z Z (C , T )
Z Z ( E , T )
(6.31)
wZ * ( F , T )
wT
wZ (C , T )
wT
wZ ( E , T )
wT
I (C , T ) I
I ( E , T ) I
(6.32)
&
q 0 ( F * , T * , X& T * )
&
q0
&
q 0 (C , T , X& T )
or
&
&
q 0 ( E , T , X& T ) q 0
(6.33)
To avoid excessive symbolism, we omit the symbols at the top part of the tensor.
As we saw in Chapter 3, P is related to the second Piola-Kirchhoff stress tensor S
(reference configuration) by means of the equation S F 1 P , then, by taking into
account (6.28) we can conclude that:
S ij
Fik1Pkj
Fik1S 0
wZ ( F , T )
wFkj
Fik1S 0
wZ (C , T ) wC pq
wC pq wFkj
wFrq
wZ (C , T ) wFrp
Frq Frp
wC pq wFkj
wFkj
S 0 Fik1
wZ (C , T ) w ( Frp Frq )
wC pq
wFkj
S 0 Fik1
wZ (C , T )
E rk E pj Frq Frp E rk E qj
wC pq
S 0 Fik1
S 0 Fik1
>
wZ (C , T )
wZ (C , T )
Fkp
Fkq Fik1
wC jq
wC pj
(6.34)
wZ (C , T ) wZ (C , T )
wZ (C , T )
wZ (C , T )
E ip
S0
wC ij
wC jq
wC pj
wC ji
wZ (C , T )
2S 0
wC ij
S 0 E iq
2S 0
wZ (C , T )
wC
S0
wZ ( E , T )
wE
(6.35)
349
Thus, the constitutive equations can be expressed in the reference configuration as follows:
Z Z ( E , T )
wZ ( E , T )
S S0
wE
thermoelastic material
wZ ( E , T )
I( E , T )
(Reference configuration)
wT
&
&
q 0 q 0 ( E , T , X& T )
(6.36)
1
P F T with which the constitutive
J
P S0
wZ ( F , T )
FT
wF
&
&
&
It is also true that q 0 Jq F T q
S
wZ ( F , T )
FT
S
S 0 0 wF
(6.37)
&
J 1 q 0 F T , (see Chapter 2). Hence we can express
wF
wZ ( F , T )
I(F , T )
wT
&
1 &
q
J q 0 ( F , T , X& T ) F T
&
J 1 F q 0 ( F , T , X& T )
(6.38)
(Current configuration)
Then, due to the principle of objectivity, the Helmholtz free energy can be written as a
function of C , i.e. Z (C , T ) . Additionally, the constitutive equation for stress
wZ (F , T )
F jk can still be rewritten as:
V ij S
wFik
V ij
wZ ( F (C ), T )
F jk
wFik
wZ (C , T ) wC pq
F jk
wC pq wFik
(6.39)
350
Z Z (U, T )
&
X
RT
U 1
Z Z (C , T )
Z Z ( E , T )
B0
R U
&
X
&
x
Z Z(F , T )
C, E
reference
configuration
b, e
current
configuration
wZ (C , T ) wC pq
F jk
wC pq wFik
wFrq
wZ (C , T ) wFrp
Frq Frp
F jk
wC pq wFik
wFik
wZ (C , T )
E ri E pk Frq F jk Frp E ri E qk F jk
wC pq
wZ (C , T )
Fiq F jp Fip F jq
wC pq
wZ (C , T ) w ( Frp Frq )
F jk
wC pq
wFik
>
>
SFiq
wZ (C , T )
wZ (C , T )
F jp SFip
F jq
wC pq
wC pq
SFiq
wZ (C , T )
wZ (C , T )
F jp SFip
F jq
wC qp
wC pq
2SFiq
@
(6.40)
wZ (C , T )
F jp
wC qp
2SF
wZ (C , T )
FT
wC
(6.41)
S0
wZ (C , T )
FT
2F
J
wC
1
F S F T
J
351
wZ (C , T )
1
FT
F 2S 0
J
wC
w: (C , T )
1
FT
F 2
J
wC
(6.42)
where S is the second Piola-Kirchhoff stress tensor, and : (C , T ) is the strain energy per
unit reference volume.
The constitutive equation for heat conduction can also be expressed as:
&
q*
&
T
J 1 q*0 ( F * , T * , X& T * ) F *
&
T
J 1Q q 0 (Q F , T , X& T ) >Q F @
&
T
J 1Q q 0 (Q R U, T , X& T ) >Q R U@
(6.43)
&
T
J 1Q q 0 (Q R U, T , X& T ) >Q R U@
&
T
J 1R T q 0 (R T R U, T , X& T ) R T R U
&
J 1R T q 0 (U, T , X& T ) U
&
J 1q 0 (U, T , X& T ) R U
&
J 1q 0 (U, T , X& T ) F
>
(6.44)
After that, so as to satisfy the principle of objectivity, the constitutive equations can be
expressed as:
Z Z(C , T )
wZ (C , T )
FT
2SF
wC
wZ (C , T )
I (C , T )
wT
&
&
q J 1 q 0 (U, T , X& T ) F
(6.45)
(Current configuration)
&
J 1 q0 (C , T , X& T ) F
Phenomenon of energy dissipation (due to the internal friction) appears when one
part of the system has a relative shearing motion with respect to other part of the
352
system, (see Romano et al. (2006)). In this way, the material response depends on
& &
the spatial velocity gradient ( x& v ( x , t ) { l F F 1 ).
Now we can observe that the functionals will also depend on the history of F :
Z(t ) Z ( F ( W ) , F ( W) , T ( W) , X& T ( W) )
I (t ) I ( F ( W) , F ( W) , T ( W) , X& T ( W) )
&
&
q 0 (t ) q 0 ( F ( W) , F ( W) , T ( W) , X& T ( W) )
(6.46)
P (t ) P ( F ( W) , F ( W) , T ( W) , X& T ( W) )
Then, to obtain the constitutive equations, we use alternative proof to that made on a
simple thermoelastic material, (see Romano et al. (2006)).
Once again we can apply the Clausius-Duhem inequality:
>
(6.47)
.
wZ wZ wZ
wZ
:F
:F
T
X& T
wF
wT
w X& T
wF
(6.48)
. 1&
wZ
wZ
wZ
wZ
&
&
P S 0 wF : F S 0 wF : F S 0 wT I T S 0 wT X T T q 0 X T t 0
or
.
&
wZ T
Tr P S 0
F S 0 wZ : F S 0 wZ I T S 0 wZ X& T 1 q 0 X& T t 0
w
w
F
T
w
T
T
w
F
(6.49)
where we have verified that given two tensors A and B , the following holds
A : B Tr ( A B T ) . Then, we can restructure the above equation as follows:
^a`^u`T
bt0
(6.50)
where
^a`
b
wZ
wZ
wZ
I ; S 0
S 0 ;S 0
w X& T
wF
wT
wZ
1&
P F T q 0 X& T
Tr S 0
wF
T
^u`
.&
F , T , X T
(6.51)
Since ^a` and b are independent of ^u` , the inequality in (6.50) holds for any arbitrary
value of ^u` , if and only if ^a` ^0` and b t 0 with which we can make the conclusion
that:
wZ
0 (The energy does not depend on F )
wF
wZ
wZ
S0
I
(The constitutive equation for entropy)
I 0
T
wT
w
S0
(6.52)
S0
wZ
w X& T
353
&
0 (The energy is not a function of the temperature gradient X& T )
Then, if we take into account the considerations in (6.52), we can rewrite the ClausiusDuhem inequality as:
wZ 1 &
&
P S 0 wF : F T q 0 X T t 0
(6.53)
We can now break down the tensor P into static and dynamic equilibrium parts, i.e.
P P ( e ) P ( d ) with which the above inequality becomes:
wZ 1 &
(e)
(d )
&
(P P ) S 0 wF : F T q 0 X T t 0
wZ
(e )
(d )
1&
&
P S 0 wF : F P : F T q 0 X T t 0
(6.54)
P (e ) ( F , F
0, T , X& T
&
wZ ( F , T )
0) S 0
wF
(6.55)
(6.56)
&
which must be met for any thermodynamic process. Note that P (d ) and q 0 cause the
energy dissipation in the system. In this way, we can summarize all the constitutive
equations for thermoviscoelastic materials as follows:
Z Z(F , T )
wZ ( F , T )
I
P (e)
P (d )
wT
thermoviscoelastic material and
wZ ( F , T )
thermodynamic constraints
S0
wF
(Reference configuration)
1&
: F q 0 X& T t 0
T
(6.57)
P (d ) is a function of P ( d ) ( F , D, W, T , X& T ) .
(6.58)
where we have considered that W *
Chapter 4).
Q T Q W Q T , and D *
Q
Q D Q T , (see
354
The equation in (6.58) must satisfy for any orthogonal tensor, including the particular case
Q T Q W Q T becomes
when Q 1 . In this situation, the term W * Q
*
W W W 0 , thus:
W Q
P ( d ) ( F , D, W, T , X& T ) P ( d ) ( F , D,0, T , X& T )
(6.59)
&
J q x& T holds, which can be proved
&
&
or q 0 i J q k Fik1 , i.e.:
&
J q F T
q0 i
wT
wX i
J q k Fik1
wT wx p
wx p wX i
J q k Fik1
wT
F pi
wx p
J q k E pk
wT
wx p
J qk
wT
wx k
J : D dV
: DdV ,
: DdV S : E dV 2 S : CdV P : F dV S
V0
V0
V0
V0
V0
P : F dV
(6.60)
Hence we can express the constitutive equations for thermoviscoelastic materials in the
current configurations as:
Z Z(F , T )
wZ ( F , T )
FT
(e) S
(6.61)
The stress (d ) in the current configuration can be obtained by comparison with the
equation in (6.42),
1
F S F T , thus:
J
(d )
1
F S ( d ) ( E , E , T , X& T ) F T
J
(6.62)
Thus:
Z Z (C , T )
wZ (C , T )
FT
wC
wZ (C , T )
I (C , T )
wT
1
(d )
(d )
F S ( E , E , T , X& T ) F T
J
q q 0 (C , E , T , X& T ) F
(e)
2SF
(6.63)
355
or
Z Z ( E , T )
wZ ( E , T )
FT
wE
wZ ( E , T )
I( E , T )
wT
1
(d )
F S ( d ) ( E , E , T , X& T ) F T
J
q q 0 ( E , E , T , X& T ) F
(e)
6.4.1
SF
(6.64)
(Current configuration)
&
q 0 ( F , T , X& T , B i )
(6.65)
P P ( F , T , X& T , B i )
where B i , i 1,2, , n , is a set of internal variables. These variables can be scalars, vectors
or higher order tensors.
In a process in which energy dissipation takes place, the theory with internal variables
together with the Clausius-Duhem inequality provides the conditions (restrictions) to the
constitutive equations.
From now on we will assume that the Helmholtz free energy is independent of the
temperature gradient, so, the Helmholtz free energy (6.65) is expressed by:
Z Z(F , T , Bi )
(6.66)
where B i ^B1 , , B n ` are the internal variables which must be added in order to
characterize the material behavior and whose presence requires that new equations be
included in the model. These additional equations are only dependent on the
thermodynamic state at the point in question, so they are local by nature.
Then, the rate of change of the equation in (6.66) is given by:
Z
wZ wZ wZ
i
B
:F
T
wB i
wT
wF
(6.67)
The operator is substituting with the number of contractions of the B order. That is, if
B is a scalar, has no contractions, if B is a vector, (scalar product), if B is a
second-order tensor, : (double scalar product) and so on.
356
Then, by combining the equation in (6.67) with the entropy inequality we obtain:
1&
wZ
wZ
wZ
i q 0 X& T t 0
P S 0 wF : F S 0 wT I T S 0 wB B
T
(6.68)
wZ
wF
I
wZ
wT
(6.69)
(6.70)
S 0
wZ
wB i
(6.71)
Then, to fully characterize the constitutive equations, the complementary laws associated
with the dissipative mechanism must be introduced, i.e. the equations for the variables
&
1&
i . One way to ensure that the equations related to q 0 and B
i satisfy the
q 0 and B
T
(6.72)
which is a convex potential for any value of the pair ( $ i , X& T ). Then, the variables are
determined by:
B
w'
w$ i
1&
q0
T
w'
w X& T
(6.73)
Problem 6.1: Find the governing equations for a continuum solid which has the following
features: Isothermal and adiabatic processes; an infinitesimal strain regime and a linear
elastic relationship between stress and strain.
b) Once the linear elastic, stress-strain relationship has been established, find the equation
in which ( ) is a tensor-valued isotropic tensor function.
Solution:
When we have isothermal and adiabatic processes temperature and entropy play no role.
In an infinitesimal strain regime, the following is satisfied:
&
Strain tensors: E | e | sym u
Stress tensors: P | S |
F | 1 ; X& | x& |
; S | S 0 . If we take this approach, mass density is no
longer unknown.
Then, taking into account the fundamental equations in (6.6)-(6.9), the remaining equations
for the proposed problem are:
1) The equations of motion
&
&
Sb Sv
2) The energy equation
&
&
&
S 0 u ( X , t ) S : E X& q 0 S 0 r ( X , t )
357
Su :
Du D
>Z TI @ Z :
or in terms of the Helmholtz free energy:
Dt Dt
SZ : e :
where : e is the energy density (also known as strain energy density). Then if we bear in mind
the entropy inequality, we can observe that the proposed problem is characterized by a
process without any energy dissipation (an elastic process), i.e. all stored energy caused by
will recover when 0 .
3) In this problem, the constitutive equations in (6.36) become:
Z Z ( )
S|
wZ ( )
w
w: e ( )
w
( )
Energy ( Z ) and stress are only functions of strain. Then, if we calculate the rate of change
wZ ( )
of the Helmholtz free energy, i.e. Z ()
: , and by substituting it with the equation
w
SZ
: e
: , we obtain:
wZ ( )
:
w
w: e ( )
:
w
:
w: e ( )
w
Thus, we can conclude that the energy equation is a redundant one, i.e. if the stress is
known the energy can be evaluated and vice-versa. So, we can summarize the governing
equations for the problem proposed with:
The equations of motion:
&
&
Sb Sv
&
(3 equations)
Su
Kinematic equations:
w: e ( )
(6 equations)
w
&
sym u (6 equations)
(6.74)
&
The unknowns of the proposed problem are: (6), u (3) and (6), making a total of 15
unknowns and 15 equations, so the problem is well-posed. Then, to achieve the unique
solution of the set of partial differential equations given in (6.74) one must introduce the
initial and boundary conditions, hence defining the Initial Boundary Value Problem for the
linear elasticity problem.
NOTE: Although the energy equation is a redundant one, at the time of establishing an
analytical or numerical method for solving the problem, we will always start from energy
principles, hence the importance of studying the energy equation in a system.
In subsection 1.6.1 The Tensor Series (Chapter 1), we saw that we can approach a tensorvalued tensor function by means of the following series:
w 2 ( 0 )
1
1 w ( 0 )
1
( 0 )
: ( 0 ) ( 0 ) :
: ( 0 )
0!
1! w
2!
w w
w ( 0 )
w 2 ( 0 )
1
: ( 0 ) ( 0 ) :
: ( 0 )
| 0
2
w
w w
Then, by considering the application point 0 0 and ( 0 ) 0 0 , and also taking
into account that the relationship - is linear, higher order terms can be discarded, thus:
( ) |
358
w ( 0 )
:
w
( )
w 2 : e ( 0 )
:
w w
Ce :
w 2 : e ( )
is a symmetric fourth-order tensor which is known as the elasticity
w w
where C e
:
1
0
: ( 0 )
: e0 0 : ( 0 ) ( 0 ) :
2
w w
w 2 : e ( 0 )
1
1
:
:
: Ce :
2
w w
2
where we have also considered that 0 0 : e0 0, 0 0 .
0!
1!
To better illustrate the problem established here, let us consider a particular case (a onedimensional case) where the stress and strain components are given by:
V ij
V 0 0
0 0 0
0 0 0
H 0 0
0 0 0 V
11
0 0 0
H ij
e
H11 V
C1111
EH
In this case, the stress-strain linear relationship becomes V EH (Hookes law) and the
strain energy density is given by : e
1
VH
2
1
HEH , (see Figure 6.4).
2
: e (H)
Current state
V(H)
:e
:e
1
HEH
2
Stored energy
1
:e
VH
2
E
V0
e
0
0
H0
: e (H ckl ) : e (H kl )
e
Then, taking into account that the relationship - is given by V ij () C ijkl
H kl (indicial
notation), we can conclude that:
e
C cijkl
H ckl
e
C ijkl
H ckl
e
C cijkl
e
C ijkl
359
Vcij
Hc22
Vc12
e
C cijkl
H ckl
Vc11
xc1
P
V cij
Hc12
a ip a jq V pq
Hc11
H 22
V ij
H12
V 22
e
C ijkl
H kl
V12
H11
x1
c
H c22
Isotropic material
e
C ijkl
e
C cijkl
ce
C cijkl
: e (H ckl ) : e (H kl )
V11
Vcijc
cc
H11
c e H cklc
C cijkl
c
V c22
P
cc
V11
x1cc
Principal space
360
Behavior of Solids
In 1660, Robert Hooke discovered that for many materials (solids) displacement was
proportional to the applied force, hence the notion of elasticity was established, but this
was not the case in the sense of the stress-strain relationship. It was the Swiss
mathematician Jacob Bernoulli who observed that the proper way to describe any change
in length was by providing a force per unit area (stress) as a function of the elongation per
unit length (strain), (see Figure 6.1).
Now, if we consider the one-dimensional stress-strain relationship seen in a loading
/unloading process we can observe the following types of behavior (see Figure 6.6):
x
Inelastic behavior.
V
loading path
loading path
Stored energy
(recoverable)
Recoverable
energy
unloading path
unloading path
Dissipated energy
V
loading path
Elastic energy
(recoverable)
unloading path
H
c) Inelastic behavior
361
completely recovered. In the linear elastic process stress-strain curve (see Figure 6.6(a)), the
paths of the loading and unloading processes are the same.
In the case of small deformations and isothermal processes, materials that behave
according to Figure 6.6(a) can be characterized by means of Linear Elasticity (Chapter 7).
A non-linear elastic process (Figure 6.6(b)) differs from a linear elastic one by the nonlinearity of the stress-strain curve. In general, materials that behave in this way tend to
show large deformations. These types of materials can be characterized by means of
Hyperelasticity (nonlinear elasticity), which is the theme in Chapter 8. It should be stressed here
that in the elastic process (whether this is linear or non-linear), the constitutive equation is
only dependent on the current value of H , i.e. it is independent of the deformation history.
In contrast to the above, inelastic behavior is characterized by involving energy dissipation,
and this energy at an atomic level can be interpreted as being the energy released for
restructuring atoms (dislocations). The tensile testing shown in Figure 6.7 is a typical
example of inelastic behavior or elastoplastic behavior to be precise. Then, materials having
these characteristics are analyzed by means of plasticity models (Chapter 9). At a macroscopic
level, elastoplastic behavior is characterized by the fact that once the stress state exceeds a
certain threshold (the yield stress) the material acquires a permanent strain, H p -plastic
strain, i.e. when the material is stress free it no longer returns to its initial state.
VY
I - elastic zone
II - plastification zone
III - completely unloading
VY
III
II
V Y - yield stress
II
H p - plastic strain
VY
H e - elastic strain
E
III
H
Hp
He
362
has no permanent strain, but internally the material will have undergone internal
degradation (an irreversible process). In Figure 6.8 we consider a loading/unloading/
loading process, where the steps 1-2-3 represent the loading process; the path 4 indicates
the unloading process which if applied will proceed as indicated by the path 5. In general,
brittle materials such as concrete, ice, ceramics and ice have these characteristics.
V
3
2
4
E
1
5
Ed
Ed E
Temperature Effect
When materials are subjected to a temperature change their mechanical properties change,
i.e. they are temperature dependent. There are two possible scenarios to be aware of when
analyzing the effect of temperature. The first is when the effect of temperature does not
significantly affect the material mechanical properties. In this case we can decouple the
problem, i.e. we can treat the thermal and mechanical effects independently. The second is
when temperature has a significant effect on the mechanical properties. In this case the
thermal and mechanical variables must be considered simultaneously in the constitutive
equations.
6.5.1.2
Tensile testing consists of a specimen whose ends are subjected to a tensile force as shown
in Figure 6.9. Then, if we known the dimensions of the specimen cross section, b , h , and
the tensile force F , it is possible to obtain the nominal stress ( V nom
F
). Furthermore,
A0
363
F
Initial cross section area
A0
'"
"
bh (undeformed area)
V|S
H
extensometer
F
A0
'"
"
364
V nom
tan
1 E
Vu
VY
E
ESH
V
E tan H
Vr
Ve
ES
E tan )
II
III
IV
Ductile Materials: large deformation occurs; there is a previous warning of failure, e.g.
as found in steel, aluminum, etc.
Depending on the manufacturing process and the amount of carbon involved in its
manufacture, some steel may behave as if it were a brittle material.
365
steel
VY
iron
concrete
VY
VY
V Y( 2 )
Ep
1
1
V Y(1)
E (p2 )
E (p1)
a) Perfectly elastoplastic
model
b) Linear hardening
elastoplastic model
c) Bi-linear hardening
elastoplastic model
V
VY
d) Softening behavior
Ea
V( Pa ) H(u10 3 )
6.67
13.3
20
24
22
0.667
1.33
2
3
3 .6
Calculate Youngs modulus ( E ) and define the stress-strain curve limit points.
366
Solution: First, we verify that the first three points maintain the same proportionalities:
E
V (1)
H (1)
V ( 2)
H ( 2)
V ( 3)
H ( 3)
20
2 u 10 3
10 000 Pa 10 kPa
The stress-strain curve can be appreciated in Figure 6.13, in which we define the following
points: V e - the proportionality point; V Y - the yield point; V u - the ultimate strength
point; and V r - the rupture strength point.
V(Pa ) 30
Vu
VY
25
20
15
Vr
3; 24
3.6; 22
2; 20
1.33; 13.3
10
0.667; 6.67
0; 0
0
0 .2%
0.5
1.5
2.5
3.5
4
3
H(u10 )
A direct test for tensile stress in brittle materials was put forward by Brazilian engineer
Fernando Lobo Carneiro. In this test a compression cylinder was used as shown in Figure
6.14, which had an advantage when working with fragile materials (concrete, ceramics),
since the manufacturing process of the specimen like the type described in Figure 6.9 can
affect the material properties.
Unconfined Compression Test
Conversely, the compression test is the opposite of the tensile test as the specimen is in the
shape of a solid circular cylinder, (see Figure 6.14). Here, the mechanical properties present
include the elastic modulus for compression, the yield stress and the rupture strength point.
In some materials, such as steel, the properties present in both the tensile and compression
tests are identical, whereas other materials, such as concrete, exhibit different properties
depending on which test is carried out. Figure 6.14 shows how concrete typically behave.
We can clearly see that it is a low tensile strength material.
Although tensile yield stress in metals is equal to compression, it was observed
experimentally that when these materials were subjected to cyclic loads their yield stress
changed. For example, let us assume that a metal which originally has as tensile and
compressive yield stresses the following values V Y and V Y , respectively, (see Figure
6.15). Once it exceeds the tensile yield stress V Y and then is subjected to an unloading
process, the compressive yield stress changes to V *Y . This phenomenon was first detected
by Bauschinger, hence it is known as the Bauschinger effect.
367
Brazilian test
V
Concrete: V Yc | 10V Yt
V Yt
Tension
Et
Compression
Ec
V Yc
VY
V *Y
VY
368
The triaxial compression test, (see Figure 6.16), is used to obtain properties of cohesive
saturated (or unsaturated) soils. A triaxial test is outlined below:
1. The specimen is a cylindrical sample.
2. The specimen is enclosed within an elastic membrane (rubber), and both ends
of the specimen are supported by rigid plates.
3. The specimen is placed in a pressure chamber and confined to pressure
denoted by V 3 .
4. A device is previously fixed to the specimen in order to measure how its length
varies and which is used to calculate strain.
F
pressure
rigid plate
flexible membrane
soil
liquid
drainage
Figure 6.16: Cross section of the triaxial compression test apparatus.
With a fixed hydrostatic pressure during the test, the normal force increases gradually until
the soil sample fails by shear. The same test is repeated by varying the hydrostatic pressure
value. Each trial stress state at the time of failure is represented by the Mohrs circle. The
envelope curve to the circles is used to define some parameters, namely: the angle of
internal friction ( G ) and cohesion ( c ), (see Figure 6.17).
NOTE: The sign convention in soil and rock mechanics is the reverse of the one adopted
in solid mechanics, i.e. in soil mechanics compression is considered to be positive since
tensile strength in soils is very low or non-existence.
Some soils are formed by sediments that are linked by electrostatic forces between fine
particles, e.g. clay-water. Negatively charged clays are cohesive with water that has a strong
electrical polarity and cohesion has the same unit of measurement as stress.
Another important parameter for cohesive materials is dilatancy ( \ ), which can be
obtained from a test as indicated in Figure 6.18.
369
specimen 3
specimen 2
specimen 1
V (I1II)
V (II2I)
( 3)
V (I1) V III
V (I3)
V (I2)
VN
tg \
dx
dy
V Pp
(6.75)
6.5.2
Behavior of Fluids
Gases as well as liquids are materials made up of molecules (agglomerations of two or more
atoms). Fundamentally, we can state that solids can resist shear stress and consequently
have the ability to store mechanical energy while liquids have low or no resistance to shear
stress and have no capacity to store energy. Additionally, resistance to shear stress in fluids
is directly linked to fluid properties, namely, viscosity and fluids are classified as nonviscous (e.g. water) or viscous (e.g. oil). In the case of viscous fluids all dissipated energy is
caused by viscosity.
370
gas (air)
Vf
V
V water
water
V solid
solid
Viscosity
W
J
N
m2 s
kgm s
s2 m2
Pa u s (6.76)
where W is shear stress, and J is the rate of change of the shear strain. The SI unit of
dynamic viscosity is (Pascal x second) Pa u s or
kg m
.
s
The most accurate way to measure dynamic viscosity is by viscometers (also called
viscosimeter), which are devices that measure the time it takes for a fluid to pass through a
very precise capillary diameter. The kinematic and dynamic viscosities are related by:
-
Iv
S
m2
(6.77)
371
6.5.3
To understand viscoelastic behavior, we can carry out a simple experiment. For example,
we can take a gum (used) and stretch it in such a way that most of the gum is concentrated
at one end. Then, we place it in a vertical position so that the only force acting on it is the
gravity, (see Figure 6.20 at time t 0 ). Without any force added to the system, we will observe
that over time the gum will start to deform, (see Figure 6.20 during time t1 o t 3 ). After it
has been deforming for a while, we cut its end off, i.e. we remove the force, and we will see
that part of the deformation recovers instantly, and we will also verify that over time
another part of the deformation recovers slowly.
That is, these materials have the ability to store mechanical energy as elastic solids and can
also dissipate energy due to their viscosity. Hence, when we are working with how to
approach the constitutive equation for these materials we have to take into account these
phenomena simultaneously, (see Findley et al. (1976), Christensen (1982)).
t0
t1
t4
t3
Instantaneous elastic
recovery
t 5 !! t 4
Slow recovery
In other words, viscoelastic materials are those in which the stress-strain relationship is
time dependent. The most relevant viscoelastic phenomena are listed below:
372
Creep When stress is constant, strain increases over time. For example we can mention a
building column, which, when force is first applied shows an initial strain, which increases
over time with no corresponding increase in stress, (see Figure 6.21).
Relaxation When strain is constant, stress decreases over time. As an example we can
cite a prestressed cable bridge whose cable is initially subjected to an initial strain causing
an initial stress and over time this stress decreases while the strain remains constant, (see
Figure 6.22).
On a final note, creep and relaxation are reciprocal phenomena.
t0
a) unloaded column
t !! t 0
b) instantaneous deformation
Figure 6.21: Creep phenomenon.
a) Unloaded cable.
b) Imposed deformation.
H0
H0
t0
t !! t 0
V e (t 0 )
V(t ) V e (t 0 )
H0
H0
6.5.4
Rheological Models
Now we can introduce some simple devices that will help us to interpret constitutive
models and which will also help us to formulate more complex constitutive models.
Let us consider a rod (a one-dimensional case) subjected to tension where the stress state at
a material point is represented by V . If we are working with a linear elastic material the
stress-strain relationship is given by V EH (Hookes law), (see Problem 6.1). If we then
compare this with the governing law of a spring given by F ku , where k is the spring
constant and u is the displacement, we can state that the linear elastic model, V EH , can
be represented by the spring device, (see Figure 6.23).
373
material point
F
Nanoscopic level
9
| 10 m
V, H
Mesoscopic level
Macroscopic level
6
| 10 3 m
| 10 m
EH
E
Spring device
V
Energy
374
VY
loading
Unloading/loading
VY
Hp
Hp
Iv
H
V I v H
Iv
1
DH
{ H
Dt
Dashpot device
Figure 6.25: Viscous dashpot device.
V
VY
VY
E
H
7 Linear Elasticity
Linear Elasticity
7.1 Introduction
Approaching the problem via the linear elasticity theory is perfectly acceptable in many
practical cases in engineering. Linear elasticity is used when the displacement gradient is
sufficiently small when compared with the unity. In this scenario we can apply the
infinitesimal strain regime (small deformation) which was discussed in Chapter 2 (see
subsection 2.14). In this approach the material strain (Green-Lagrange) and the spatial
strain tensor (Almansi) collapse into:
>
&
&
1
(u) (u) T
2
&
( x, t )
&
sym u
H ij
1 wu i wu j
2 wx j wx i
1
u i , j u j ,i
2
(7.1)
&
H ij
H11
H
12
H13
H 12
H 22
H 23
H13
H 23
H 33
wu1
wx1
1 wu1 wu 2
2 wx 2 wx1
1 wu
wu
1 3
2 wx3 wx1
1 wu1 wu 2
2 wx 2 wx1
wu 2
wx 2
wu
u
w
1
2 3
2 wx3 wx 2
1 wu1 wu 3
2 wx3 wx1
1 wu 2 wu 3
2 wx3 wx 2
wu 3
wx3
(7.2)
Then, taking into account the following nomenclature used in engineering notation:
displacement: u1 u , u 2 v , u 3 w , and the strain field: H11 H x , H 22 H y , H 33 H z ,
2H 12 J xy , 2H 23 J yz , 2H 13 J xz , the equations in (7.2) can be rewritten as follows:
E.W.V. Chaves, Notes on Continuum Mechanics, Lecture Notes on Numerical
Methods in Engineering and Sciences 4, DOI 10.1007/978-94-007-5986-2_9,
International Center for Numerical Methods in Engineering (CIMNE), 2013
375
376
H ij
Hx
1
2 J xy
1 J xz
2
1
2
J xy
Hy
1
2
J yz
1
2
1
2
J xz
J yz
H z
wu
wx
1 wu wv
2 wy wx
1 wu ww
2 wz wx
1 wu wv
2 wy wx
wv
wy
1 wv ww
2 wz wy
1 wu ww
2 wz wx
1 wv ww
2 wz wy
ww
wz
(7.3)
Su
x3
dV
& &
t * ( x)
& &
Sb( x )
&
x
x2
x1
7.2.1
Governing Equations
In Problem 6.1 we established the governing equations for the linear elasticity problem,
i.e.:
& &
& &
&
w 2 u( x, t )
( x , t ) S b( x , t ) S
wt 2
wV ij
wx j
Sb i
w 2ui
wt 2
(7.4)
C:
V ij
C ijkl H kl
(7.5)
7 LINEAR ELASTICITY
377
H ij
1 wu i wu j
2 wx j wx i
(7.6)
7.2.2
&
&
ui ( x, t ) u i * ( x, t )
(7.7)
V jk n k
&
t j * ( x, t )
(7.8)
&
& &
u( x , t 0) u 0
& &
& &
wu 0 ( x , t )
& &
u 0 ( x, t ) v 0 ( x )
wt
t 0
&
&
u i ( x , t 0) u 0 i ( x )
&
u 0 i ( x ) v 0 i
(7.9)
In the particular case when we have a static or quasi-static problem, the equation of motion
& &
becomes the equilibrium equations ( Sb 0 ), and the initial conditions become
redundant.
C:
Indicial notation
V ij
C ijkl H kl
(7.10)
where C is known as the elasticity tensor (or elastic stiffness tensor), which is of the symmetric
fourth-order type and contains the elastic constants (the material properties). In Problem
6.1 we showed that C has both minor ( C ijkl C jikl C ijlk ), and major ( C ijkl C klij )
symmetry, so the tensor features 21 independent components. It is said that a material is
homogeneous when its elastic properties do not vary from point to point throughout the
continuum, i.e. C is independent of the position vector. Moreover, a material is said to be
isotropic at any point when the components of C do not change if the reference system
undergoes a base change.
378
7.3.1
By referring to the Cauchy stress tensor symmetry we can use Voigt notation (see Chapter
1) to store the tensor components as follows:
V ij
V11
V
12
V
13
V13
Voigt
o^T`
V 23
V 33
V12
V 22
V 23
V11
V
22
V 33
V12
V 23
V13
V x
V
y
V z
W xy
W yz
W xz
V xx
V
yy
V zz
V xy
V yz
V xz
(7.11)
Engineering
Notation
Each Cauchy stress tensor component ( V11 , V 22 , V 33 , V12 , V 23 , V13 ) can be obtained by
using the constitutive equation in (7.10). Then by expanding said equation for the
component V11 , we find:
V11
C11kl H kl V11
C1122 H 22
C1132 H 32
C1113 H13
C 1123 H 23
C 1133 H 33
H lk l C ijkl
(7.12)
becomes V11 C1111 H11 C1122 H 22 C1133 H 33 2C1112 H12 2C1123 H 23 2C1113 H13 .
Likewise, we can obtain the expressions for V 22 , V 33 , V12 , V 23 and V13 , and if we then
reorder them in matrix form we can obtain the generalized Hookes law in Voigt notation,
i.e.:
V11
V
22
V 33
V12
V 23
V13
C1111
C
2211
C 3311
C 1211
C 2311
C 1311
C1122
C1133
C1112
C1123
C 2222
C 3322
C 2233
C 3333
C 2212
C 3312
C 2223
C 3323
C1222
C 2322
C1233
C 2333
C 1212
C 2312
C1223
C 2323
C1322
C1333
C 1312
C1323
C1113 H11
C 2213 H 22
C 3313 H 33
C1213 2H 12
C 2313 2H 23
C1313 2H 13
^T` >C @ ^ `
(7.13)
where >C @ is the matrix with mechanical elastic properties. Then by application of major
symmetry, i.e. C ijkl C klij , the elasticity tensor components can be expressed in Voigt
notation as follows:
>C @
C11
C
12
C13
C14
C15
C16
C12
C13
C14
C15
C22
C23
C23
C33
C24
C34
C25
C35
C26
C36
C46
C56
C24
C25
C34
C35
C44
C45
C45
C55
C16
C26
C36
C46
C56
C66
(7.14)
7 LINEAR ELASTICITY
379
The equation in (7.13) indicates that the strain tensor components in Voigt notation are
shown in the following format:
H 11
H
12
H
13
H ij
H 12
H 22
H 23
H 13
Voigt
o^ `
H 23
H 33
H xx
H
yy
H zz
2H xy
2H yz
2H xz
H 11
H
22
H 33
2H12
2H 23
2H13
Hx
H
y
Hz
J xy
J yz
xz
J
(7.15)
Engineering
Notation
Note that as the double scalar product : has the unit of stored energy (see Problem
6.1), then, this energy must be the same as when it is obtained either by : or when
^T`T ^ `, i.e.:
1
:
,
2 Tensorial
7.3.2
1
V11 H11 V 22 H 22 V 33 H 33 2V12 H12 2V 23 H 23 2V13 H13
2
1
T
^T`
^
`
2
Voigt
Law
for
(7.16)
the
Voigt notation
^T` >C @ ^ `
C ijkl H kl
(7.17)
These components are affected by any change to the coordinate system. Then, given the
new coordinate system xc1 , xc2 , xc3 , the generalized Hookes law therein is given by:
Indicial notation
Vcij
Voigt notation
^T c` >C c@ ^ c`
C cijkl Hckl
(7.18)
where Vcij , Hckl and C cijkl show the stress, the strain, and the elasticity tensor components
in the new system xc1 , xc2 , xc3 respectively which are explicitly given in Voigt notation, by:
^ c`
and
Hc11
Hc
22
Hc33
c
2H 12
2Hc23
2Hc13
Hcxx
Hc
yy
Hczz
2H cxy
2Hcyz
2Hcxz
Hcx
Hc
y
Hcz
J cxy
Jcyz
Jcxz
^Tc`
Vc11
V c
22
Vc33
c
V12
Vc23
Vc13
Vcxx
V c
yy
Vczz
Vcxy
Vcyz
Vcxz
Vcx
Vc
y
Vcz
Wcxy
Wcyz
Wcxz
(7.19)
380
>C c@
C11c
C c
12
C13c
C14c
C15c
C16c
C12c
c
C22
c
C23
c
C24
c
C25
c
C26
C13c
c
C23
c
C33
c
C34
c
C35
c
C36
C14c
c
C24
c
C34
c
C44
c
C45
c
C46
C15c
c
C25
c
C35
c
C45
c
C55
c
C56
C16c
c
C26
c
C36
c
C46
c
C56
c
C66
(7.20)
We will next establish the transformation law for these tensor components in Voigt
notation.
7.3.2.1
At a given point in the continuum, the stress and the strain tensor components related to
the system x1 , x 2 , x3 , are represented by V ij and H ij , respectively. Then, the components
of these second-order tensors ( Vcij , Hcij ), in the new system, can be obtained as follows:
Indicial notation
Vcij
Matrix notation
Indicial notation
H cij
(7.21)
Tc A T A T
a ik a jl V kl
Matrix notation
a ik a jl H kl
(7.22)
c A AT
a11
a12
a13
a 31
a 32
a33
A a 21 a 22 a 23
(7.23)
In Voigt notation the transformation laws in (7.21) and (7.22), (see Chapter 1), are given
respectively by:
1
inverse
^T c` >M@ ^T`
o^T` >M@ ^T c`
(7.24)
1
^ c` >N @^ ` inverse
o^ ` >N @ ^ c`
(7.25)
where >M@ is the transformation matrix of the stress tensor components in Voigt notation,
which is given explicitly by:
>M@
a11 2
2
a 21
a 2
31
a 21 a11
a a
31 21
a 31 a11
a12 2
a13 2
a 22
a 32 2
a 23
a 33 2
a 22 a12
a 32 a 22
a 32 a12
a13 a 23
a 33 a 23
a 33 a13
2a11 a12
2a 21 a 22
2a 31 a 32
2a12 a13
2a 22 a 23
2a 32 a 33
a13 a 21 a11 a 23
a 33 a 21 a 31 a 23
a 33 a11 a 31 a13
2a11 a13
2a 21 a 23
2a 31 a 33
(7.26)
and >N @ is the transformation matrix of the strain tensor components in Voigt notation:
7 LINEAR ELASTICITY
>N @
a11 2
2
a 21
2
a
31
2a 21 a11
2a a
31 21
2a 31 a11
381
a12 2
a13 2
a11 a12
a12 a13
a 23 2
a 33 2
a 21 a 22
a 31 a 32
a 22 a 23
a 32 a 33
a 22
a 32 2
2a 22 a12
2a13 a 23
2a 32 a 22
2a 33 a 23
2a 32 a12
2a 33 a13
a13 a 21 a11 a 23
a 33 a 21 a 31 a 23
a 33 a11 a 31 a13
a11 a13
a 21 a 23
a 31 a 33
(7.27)
Additionally, it can be shown that >M@ and >N @ are not orthogonal matrices, i.e.
>M@1 z >M@T and >N @1 z >N @ T , and that:
7.3.2.2
The Transformation
Components
(7.28)
Matrix
of
the
Elasticity
Tensor
^T` >C @ ^ `
^ `
(7.29)
where >Cc@ is the elasticity matrix in the new system ( xc1 , xc2 , xc3 ). Therefore, we can define
the transformation law of the elasticity tensor components in Voigt notation as:
(7.30)
Materials in general are anisotropic, i.e., material properties have different values for
different directions at any given point. Certain kinds of these at the microscopic and
mesoscopic scale have anisotropic properties, such as: concrete (made by mixing different
materials), but at a macroscopic level these properties can be considered as an average of
these properties at the mesoscopic scale, so, it is possible to consider material as
macroscopically isotropic, i.e., material properties are independent of the coordinates
system adopted. There are materials such as wood or man-made materials, e.g. composite
materials, which are made up of fibers that are directionally oriented and embedded in a
matrix, hence these materials exhibit a clear anisotropy even at the macroscopic level.
382
Most materials have some kind of symmetry along one or more axes, i.e. these axes can be
reversed without changing the material properties. For example, Figure 7.2(b) shows one
plane of symmetry, the plane x1 x 2 , which implies that if we are dealing with just one of
these we can change a coordinate system from say x1 , x 2 , x3 to xc1 , xc2 , xc3 without
altering the elastic properties of the material. Then we can see another example of
symmetry in Figure 7.2(c) in which two planes of symmetry are displayed, namely: x1 x 2
and x 2 x 3 . Remember that the transformation law from x1 - x 2 - x3 to xc1 - xc2 - xc3 system is
given by the equation:
x1c
xc
2
x 3c
a11
a
21
a 31
a12
a 22
a 32
a13 x1
a 23 x 2
a 33 x 3
(7.31)
Next, we will study the different types of symmetry that appear in materials which may
include: one plane of symmetry (monoclinic symmetry), two planes of symmetry
(orthotropic symmetry), tetragonal symmetry, transversely isotropic symmetry, cubic
symmetry, and finally symmetry in all orientation (isotropy).
xc3
x 2 , xc2
x2
x1 , xc1
x1
x3
x 2 , xc2 , x 2cc
x1cc
x1 , xc1
x3
x3
a) Original
coordinate
system
x3cc
b) One plane of
symmetry
c) Two planes
of symmetry
7.4.2
7.4.2.1
Triclinic materials are the most generic of anisotropic materials, i.e. there are no symmetry
planes. Then, the elasticity tensor features 21 independent components to be determined in
the laboratory:
>C @
C11
C
12
C13
C14
C15
C16
C12
C22
C23
C24
C25
C26
C13
C23
C33
C34
C35
C36
C14
C24
C34
C44
C45
C46
C15
C25
C35
C45
C55
C56
C16
C26
C36
C46
C56
C66
Triclinic materials
21 independent components
(7.32)
NOTE: The main drawback when dealing with high material anisotropy is the extreme
complexity that appears at the time of obtaining the constants (material properties) in the
laboratory.
7 LINEAR ELASTICITY
7.4.2.2
383
Let us now consider a material that has a single plane of symmetry (plane x1 x 2 ) as
illustrated in Figure 7.2(b). Then, the transformation law between the systems defined in
Figure 7.2(a) and Figure 7.2(b) is given by:
x1c
xc
2
x 3c
1 0 0 x1
0 1 0 x
2
0 0 1 x 3
(7.33)
with which we can obtain the transformation matrix ( >M@ ), previously defined in (7.26),
as:
>M@
1
0
0
0
0
1
0
0
0
0
0
0
1
0
0
0
0 0 0
0 0 0
0 0 0
1 0 0
0 1 0
0 0 1
(7.34)
Then, to obtain the elasticity matrix in this new system, we can carry out the following
matrix operation:
(7.35)
>C c@
C11
C
12
C13
C14
C15
C16
C12
C22
C23
C24
C25
C26
C13
C14
C23
C33
C34
C35
C36
C15
C24
C34
C25
C35
C46
C56
C44
C45
C45
C55
C16
C26
C36
C46
C56
C66
(7.36)
Since in this specific transformation, the elasticity matrix must provide symmetry, i.e.
>C c@ >C @ , we can draw the conclusion that the terms in which negative signs appear should
be zero so as to satisfy the symmetry condition. Then for materials that exhibit one plane
of symmetry, the elasticity matrix has 13 independent components, namely:
>C @
C11
C
12
C13
C14
0
C12
C22
C23
C24
0
0
C13
C23
C33
C34
0
0
C14
C44
0
0
0
C55
C56
C24
C34
0
0
0
0
0
0
C56
C66
Monoclinic Symmetry
13 independent constants
(7.37)
384
7.4.2.3
We will start from monoclinic symmetry to define the elasticity matrix format for a material
with two planes of symmetry. Then, by means of the transformation law between the
systems defined in Figure 7.2(b) and Figure 7.2(c) we obtain:
1 0 0
o>M@
1 0
0 0 1
A 0
1
0
0
0
0
1
0
0
0
0
0 0
0 0
1 0
0 1
0 0
0 0
0 0
0 0
0 0
0 0
1 0
0 1
(7.38)
where we have considered the equation in (7.26) in order to evaluate the matrix >M@ .
Then, to obtain the elasticity tensor components in the system x cc , we can carry out the
following matrix operation >C cc@ >M@ >C c@ >M@T the result of which is:
>C cc@
C11
C
12
C13
C14
0
C12
C13
C14
C44
0
0
C55
C24
0
C23
C33
C34
0
C24
C34
C22
C23
0
0
C56
0
0
0
0
C56
C66
(7.39)
In this specific transformation the following must be satisfied >C cc@ >C c@ >C @ , with which
we can draw the conclusion that the elasticity matrix features 9 independent constants to
be determined:
>C @
C11
C
12
C13
0
0
C12
C22
C23
0
0
0
C13
0
0
0
0
0
0
0
C44
0
0
0
C55
0
C23
C33
0
0
0
0
0
C66
Orthotropic Symmetry
9 independent constants
(7.40)
NOTE: Materials such as bones present a high degree of anisotropy. Nevertheless, some
researchers consider two planes of symmetry (orthotropic symmetry) to simulate
numerically the bone behavior.
7.4.2.4
Tetragonal Symmetry
Materials with tetragonal symmetry have 5 planes of symmetry one of which is the x1 x 2
plane and the other 4 are indicated in Figure 7.3. Note that this type also includes
orthotropic symmetry. So, to determine the format in which the matrix >C @ is presented we
start from the elasticity matrix given in (7.40). Next, we apply the symmetry condition
according to the transformation from x1 x 2 x 3 to x1c x 2c x3c x3 , in which the
transformation matrix is given by:
7 LINEAR ELASTICITY
0.5 0.5 0 1
0.5 0.5 0 1
0
0 1 0
0
.
5
0
.5 0 0
0
0 0 0
0 0 0
0
cos( S4 ) sin( S4 ) 0
S
S
o>M@
sin( 4 ) cos( 4 ) 0
0
0
1
x3
385
0
0
0
0
1
2
1
2
0
0
0
0
12
(7.41)
x2
xc2
xc1
S/4
S/4
S/4
S/4
S/4
S/4
S/4
S/4
x1
C13 C23
C22 C11
C C22 2 C12
0
0
C44
C44 11
4
4
2
4
C13 C23
C22 C11
C11 C22 2C12
C11 C22 2 C12
0
0
C44
C44
4
4
2
4
C13 C23
C23 C13
C13 C23
0
0
C
33
2
2
2
C22 C11
C23 C13 C11 C22 2C12
C22 C11
0
0
4
4
2
4
C
C
C
C
55
66
55
66
0
0
0
0
2
2
22
C c
(7.42)
Note that the plane ( x1c x3 ) is also a plane of symmetry, (see Figure 7.3), so, the
components of (7.42) must be equal to the components obtained from the following
coordinate transformation:
386
1 0 0
0 1 0
o>M@
0 0 1
1
0
0
0
0
1
0
0
0
0
0 0 0
0 0 0
1 0 0
0 1 0
0 0 1
0 0 0
0
0
0
0
0
(7.43)
Then, by once again applying the transformation seen in (7.30), we can obtain the
following matrix:
C11 C22 2C12
C13 C23
C11 C22
C C22 2C12
0
0
C44
C44 11
4
4
2
4
C13 C23
C11 C22
C11 C22 2C12
C11 C22 2C12
0
0
C44
C44
4
4
2
4
C13 C23
C13 C23
C13 C23
C
0
0
33
2
2
2
C
C
C
C
C
C
C
C
C
2
22
22
11
11
13
23
11
22
12
0
0
4
4
2
4
C
C
C
C
55
66
66
55
0
0
0
0
2
2
22
C c
(7.44)
and by comparing the two matrices given in (7.42) and in (7.44), we can conclude that
C11 C22 , C55 C66 , C13 C23 . Then, the elasticity matrix for tetragonal symmetry material
features 6 independent constants to be determined:
>C @
7.4.2.5
C11
C
12
C13
0
0
C12
C11
C13
0
0
0
C13
C13
C33
0
0
0
0
0
0
0
C44
0
0
0
C55
0
0
0
0
0
0
C55
Tetragonal symmetry
6 independent constants
(7.45)
Material with transversely isotropic symmetry already includes orthotropic symmetry, (see
Eq. (7.40)). In addition, any transformation on the plane x1 x 2 is also a plane of
symmetry, (see Figure 7.4).
For these kinds of material x1 x 2 and x 2 x 3 are planes of symmetry, i.e. they have
orthotropic symmetry, so, by starting from the elasticity matrix for orthotropic symmetry in
(7.40) and by some transformations on the plane x1 x 2 we can obtain the constants.
Initially, let us consider a transformation on the plane x1 x 2 characterized by the angle
B 90 , (see Figure 7.4), with which we can obtain the following transformation matrices:
7 LINEAR ELASTICITY
0 1 0
1 0 0
o>M@
0 0 1
0
1
0
0
1
0
0
0
0
0
387
0 0
0 0
0 0
0 0
0 1
1 0
0 0
0 0
1 0
0 1
0 0
0 0
x3 , x3c
(7.46)
Transformation matrix:
x 2c
x2
cos(B ) sin(B ) 0
sin(B ) cos(B ) 0
0
0
1
x1c
B
x1
>C c@
C22
C
12
C23
0
0
C12
C23
0
0
0
0
0
0
0
0
C44
0
0
C66
C11
C13
C13
C33
0
0
0
0
0
C55
(7.47)
C22 , C23
C13 , C55
C66 .
2
1
0
2
1
2
0
0
o>M@
0.5 0.5
0.5 0.5
0
0
0
.
5
0
.5
0
0
0
0
0 1
0 1
0
0
0
0
0
0
1
2
1
2
0
0
0
0
12
(7.48)
>C c@
C13
1
2
1
2
0
0
C13
C13
C13
0
C33
0
0
0
0
0
1
2
0
(C11 C12 )
0
0
0
0
0
0
C55
0
0
0
0
0
C55
(7.49)
388
Afterwards, if we compare (7.49) with (7.47) we can draw the conclusion that
C44 12 (C11 C12 ) , hence the matrix >C @ features 5 independent constants. Note that, any
other transformation on the plane x1 x 2 will not reduce the number of constants. Thus,
matrices with the elastic mechanical properties for transversely isotropic symmetry material
appear in the following format:
>C @
C11
C
12
C13
0
0
7.4.2.6
C12
C11
C13
0
0
0
C13
C13
C33
1
2
0
0
0
0
0
0
0
0
0
0
0
0
C55
(C11 C12 ) 0
0
C55
0
0
(7.50)
5 independent constants
Cubic Symmetry
Some metals are formed by crystals which can be classified as cubic symmetry materials.
These exhibit two planes of symmetry (orthotropic symmetry) and also have the same
properties if we make a rotation along the x3 -axis at an angle B 90 , and along the axis
xc1 with C 90 , as shown in Figure 7.5.
Rotation along x3 -axis
B 90
x3
C 90
x3c
x2
x 2cc
x1c
x 2c
x1
x1cc
x3cc
1 0
0 1
A 1 0 0 o>M@
0
1
0
0
1
0
0
0
0
0
0 0
0 0
1 0
0 1
0 0
0 0
0 0
0 0
0 0
0 0
0 1
1 0
(7.51)
7 LINEAR ELASTICITY
C22
C
12
C32
0
0
>C @
C21
C23
0
0
0
0
0
0
0
0
C44
0
0
C55
C11
C31
C13
C33
389
0
0
0
0
0
C55
(7.52)
Then, by comparing the above matrix with (7.40) we can conclude that:
C11
C
12
C13
0
0
>C @
C12
C13
0
0
0
0
0
0
C44
0
0
C55
C11
C13
C13
C33
0
0
0
0
0
0
0
C55
(7.53)
Then using the equation in (7.53), we rotate the xc1 -axis at an angle C 90 , resulting in the
transformation matrix:
0
o>M@
0 1
0 1 0
0
A 0
1
0
0
0
0
0
1
0
0
0
0 0
0
1 0
0
0 0
0
0 0
0
0 0 1
0 1 0
0
0
0
1
0
(7.54)
After that, if we substitute (7.54) into the transformation law (7.30), we can obtain the
elasticity matrix >Cc@ :
>C c@
C11
C
13
C12
0
0
C13
C33
C13
0
C12
C13
C11
0
0
0
0
0
0
C55
0
0
0
0
C55
0
0
0
0
0
0
C44
(7.55)
Additionally, by comparing (7.55) with (7.53) we can conclude that C33 C11 ; C55 C44 ;
C13 C12 , since >C c@ >C @ . Then, the elasticity matrix is defined by three independent
constants:
>C @
C11
C
12
C12
0
0
C12
C11
C12
0
0
0
C12
C12
C11
0
0
0
0
0
0
0
0
C44
0
0
0
0
C44
0
0
0
0
0
0
C44
Cubic Symmetry
3 independent constants
(7.56)
390
7.4.2.7
Finally, if the material has the same property in all directions it is called isotropic material.
Note that if we compare (7.56) with (7.52) we can conclude that C44 C55 12 (C11 C12 ) to
fulfill symmetry in all direction. Then, the elasticity matrix features 2 elastic constants to be
determined:
>C @
e
C11
C
12
C12
0
0
C12
C11
C12
0
0
1
(
C
11 C12 )
2
0
0
C12
C12
C11
0
1
2
1
(C11 C12 )
2
0
(C11 C12 )
0
>C @
e
1
2
Isotropic Symmetry
2 independent
constants
(7.57)
M
M
0 0 0
M 2N
M
M
N
M
2
0 0 0
M
M
M 2N 0 0 0
N 0 0
0
0
0
0
0
0
0 N 0
0
0
0 0 N
0
(7.58)
> @
where the constants M , N , are known as the Lam constants. We then split the matrix C e
as follows:
>C @
e
1 0 0 0 0 0
1 1 1 0 0 0
0 1 0 0 0 0
1 1 1 0 0 0
0 0 1 0 0 0
1 1 1 0 0 0
M
2N
1
0 0 0 2 0 0
0 0 0 0 0 0
0 0 0 0 1 0
0 0 0 0 0 0
2
1
0 0 0 0 0 0
0 0 0 0 0 2
1
1
1
>1 1 1 0 0 0 @
0
0
0
(7.59)
Note that I is the matrix in Voigt notation that represents the symmetric fourth-order
>
1
E ik E jl E il E jk ), (see Chapter 1). Then the tensor with
2
the elastic properties for isotropic materials is represented in tensorial and indicial notations
as follows:
The elasticity tensor
for isotropic material
Tensorial notation
Ce
M1 1 2NI
Indicial notation
e
C ijkl
ME ij E kl N>E ik E jl E il E jk @
(7.60)
7 LINEAR ELASTICITY
391
NOTE: Remember that in Chapter 1 we saw that any isotropic fourth-order tensor can be
written in terms of the following tensors: E ij E kl , E ik E jl , E il E jk , i.e.:
C ijkl
a 0 E ij E kl a1 E ik E jl a 2 E il E jk
(7.61)
>C @
e 1
1 MN
N (3M 2N)
M
1
2N (3M 2N)
1
M
2N (3M 2N)
M
1
2N (3M 2N)
1 MN
N (3M 2N)
M
1
2N (3M 2N)
M
1
2N (3M 2N)
M
1
2N (3M 2N)
1 MN
N (3M 2N)
> @
>C @
e 1
1
N
0
1
N
(7.62)
as follows:
1 0 0 0 0 0
1 1 1 0 0 0
0 1 0 0 0 0
1 1 1 0 0 0
1 1 1 0 0 0 1 0 0 1 0 0 0
M
2N(3M 2N) 0 0 0 0 0 0 2N 0 0 0 2 0 0
0 0 0 0 2 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 2
1
1
1
>1 1 1 0 0 0 @
0
0
0
(7.63)
I1
If we can verify firstly, that the second matrix of (7.63) is the Voigt notation representation
of the inverse of the symmetric fourth-order unit tensor components, and secondly,
1
I ijkl I ijkl
holds, then the inverse of the isotropic elasticity tensor is given as follows:
Tensorial notation
Ce
1
1
M
1 1
I
2N(3M 2N)
2N
1
Indicial notation
1
e
ijkl
ME ij E kl
2N(3M 2N)
>
1
E ik E jl E il E jk
4N
(7.64)
where C e is known as the elastic compliance tensor. Here, we will left the reader work out
1
whether C e : C e
I sym { I .
392
Constitutive Equations
The generalized Hookes law (7.10) for isotropic linear elastic materials can be written using
the equation in (7.60) as follows:
Ce :
M1 1 2NI : M1 1,
: 2N
Tr ( )
I,
:
MTr ( )1 2N
sym
(7.65)
thus
Tensorial notation
Indicial notation
V ij
MTr ( )1 2N
MH kk E ij 2NH ij
(7.66)
Indicial notation
MH kk E ij 2NH ij
MTr ( )1 2N
V ij
2N
2NH ij
MTr ( )1
M
1
Tr ( )1
2N
2N
H ij
V ij MH kk E ij
M
1
V ij
H kk E ij
2N
2N
(7.67)
We now need to evaluate the trace H kk and to do so we need to obtain the trace of V ij , i.e.:
Tensorial notation
Indicial notation
MH kk E ij 2NH ij
: 1 MTr ( )1 : 1 2N : 1
V ij
Tr ( ) MTr ( )3 2NTr ( )
V ii
1
Tr ( )
(3M 2N)
V kk
Tr ( )
H kk
MH kk E ii 2NH ii
3M 2N H kk
MH kk 3 2NH kk
(7.68)
1
V kk
(3M 2N)
Then, by substituting the Tr ( ) value given in (7.68) into the equation in (7.67) we obtain:
Tensorial notation
M
1
Tr ( )1
2N(3M 2N)
2N
Indicial notation
H ij
1
M
V kk E ij
V ij
2N(3M 2N)
2N
(7.69)
Furthermore, the above equation could easily have been obtained by means of the
following relationship:
1
Ce :
1
M
1 1
I :
N
M
N
N
2
(
3
2
)
2
M
1
1 Tr ( )
2N
2N(3M 2N)
(7.70)
where we have applied the compliance elasticity tensor given in (7.64). Then, from the
definition of the Cauchy stress tensor eigenvalues and eigenvectors, i.e. n J n , we can
obtain:
7 LINEAR ELASTICITY
393
n J n MTr ( )1 2N n J n
MTr ( )1 n 2N n J n
MTr ( )n 2N n J n
2N n J n MTr ( )n
J MTr ( )
n
n
2N n J MTr ( ) n
2N
J MTr ( )
n
n
2N
n J n
(7.71)
Thus, as expected, we can see that the tensors and have the same principal directions
J MTr ( )
.
(eigenvectors), and their eigenvalues (principal values) are connected by J
2N
If we denote by J (1) H1 , J (2) H 2 , J (3) H 3 and J (1)
eigenvalues of and can also be evaluated as follows:
7.5.2
7.5.2.1
V1
0
0
V2
H 1
0
0
H2
0
1 0 0
H1
0 MTr ( ) 0 1 0 2N 0
V 3
0 0 1
0
0
0
H 3
V 1
1
0
2N
0
V1 , J (2 )
V 2 , J (3)
0
0
H 3
0
H2
0
V 3 , the
(7.72)
0
1 0 0
MTr ( )
0
0 1 0
2N
0 0 1
0 V 3
1 0 0
V 1
M
1
Tr ( ) 0 1 0
0
2N(3M 2N)
2N
0 0 1
0
0
V2
0
V2
0
0
0
V 3
(7.73)
The normal stresses ( V x , V y , V z ) can not produce angular shear with respect to
the same coordinate system;
Pure shear stress produces deformation only in the plane where shear is applied.
Based on these assumptions, we can conclude that the strain functions are:
Hx
J xy
H x (V x , V y , V z )
Hy
J xy (W xy )
J yz
H y (V x , V y , V z )
J yz (W yz )
Hz
J zx
H z (V x , V y , V z )
J zx (W zx )
(7.74)
We can also assume that the normal strain is a linear function of the normal stresses, i.e.:
Hx
B 1V x B 2 V y B 3 V z
(7.75)
Then, because of material isotropy, the effect of V y upon H x is the same as the effect of
V z upon H x , (see Figure 7.6(a)), thus:
B2
B3 0
(experimentally observed)
(7.76)
394
with
B1
1
E
B2
B3
Q
E
(7.77)
where E is Youngs modulus (unit of stress) and Q is Poissons ratio (dimensionless). The values
of E and Q can then be obtained by laboratory experiments, (see Chapter 6). Afterwards,
all of these values will enable us to write the first three equations in (7.74) as, (see Figure 7.6):
>
1
Vx Q Vy Vz
E
Hx
7.5.2.2
>
1
V y Q V x V z
E
; Hy
; Hz
>
1
Vz Q Vx V y
E
(7.78)
The relationship between the types of shear stress and shear strain are given by:
J xy
1
W xy
G
1
W yz
G
J yz
J zx
1
W zx
G
(7.79)
where G is known as the shear modulus, which is related to the parameters E and Q by the
equation:
G
E
2(1 Q)
(7.80)
Then, to define the bulk modulus ( N ) we can start from the strain tensor trace:
I
Hx Hy Hz
>
1
V x Q V y V z V y Q V x V z V z Q V x V y
E
(1 2Q)
Vx Vy Vz
E
3(1 2Q)
Vm
E
>
@
(7.81)
In infinitesimal strain theory, (see Chapter 2), the strain tensor trace is equal to the volume
ratio, i.e.:
Hv
'V
dV0
Hx Hy Hz
(7.82)
Then, if we compare the equation in (7.81) with (7.82) we can draw the conclusion that:
Hv
3(1 2Q)
Vm
E
Vm
p
E
Hv
3(1 2Q)
(7.83)
3(1 2Q)
is the compressibility factor. We can then define the bulk modulus (or
E
compressibility modulus) ( N ), (see Figure 7.7) as the inverse of the compressibility factor:
where
E
3(1 2Q)
(7.84)
If we then look at said compressibility factor, (see equation (7.83)), when working with
3(1 2Q )
0 , which is equivalent to
E
Q 0.5 . Then, material in which Poissons ratio equals Q 0.5 is considered to be
7 LINEAR ELASTICITY
395
y
QV x
E
a)
Vx
E
Vx
H x
H y
H z
Vx
QV x
E
Vx
E
QV
x
E
QV x
E
z
y
Vy
b)
QV y
E
Vy
E
H x
H y
H z
QV y
E
QV y
E
Vy
E
QV y
E
Vy
H x
H y
H z
Vz
c)
QV z
E
Vz
E
Vz
QV z
E
QV z
E
Vz
E
=
QV z
E
H x
H y
H z
>
1
Vx Q V y Vz
E
1
V y Q V x V z
E
1
Vz Q Vx V y
E
>
>
396
W xy
Vx
y
p
Vx
W xy
p
p
E -Youngs modulus
N -Bulk modulus
G -Shear modulus
Vx
p
W xy
H v -volumetric
strain
N G
Hx
Hv
J xy
Hx
H
y
Hz
J xy
J yz
J zx
where G
V x
V
y
Vz
W xy
W yz
W zx
1
E
Q
E
Q
E
Q
E
1
E
Q
E
Q
E
Q
E
1
E
1
G
1
G
0 V x
V
y
0 V
z
W
xy
0
W yz
0 W zx
(7.85)
E
. Then, the inverse of (7.85) is given by:
2(1 Q )
Q
Q
0
1 Q
Q 1 Q
Q
0
Q
Q 1 Q
0
1 2Q
E
0
0
0
2
(1 Q)(1 2Q)
0
0
0
0
0
0
0
0
0
0
0
0
1 2Q
2
0
Hx
Hy
0 Hz
J xy
0 J yz
1 2Q J zx
2
0
0
Now, the above relationship can be rewritten in tensorial or indicial notations as:
(7.86)
7 LINEAR ELASTICITY
Tensorial notation
397
Indicial notation
Q E Tr ( )
E
1
(1 Q )(1 2Q )
(1 Q)
V ij
QE
E
H kk E ij
H ij
(1 Q)(1 2Q)
(1 Q)
(7.87)
If we then compare the equations in (7.87) and (7.66) we can conclude that the Lam
constants ( M , N ) are connected to the parameters ( Q , E ) via the following relationships:
M
QE
(1 Q )(1 2Q )
N G
E
2(1 Q)
(7.88)
Q
1 Q
Tr ( )1
E
E
Indicial notation
H ij
Q
1 Q
V kk E ij
V ij
E
E
(7.89)
Then if we compare the equations in (7.89) with those in (7.69) we can draw the conclusion
that:
E
N(3M 2N)
MN
M
2(M N)
(7.90)
Table 7.1 provides us with the different equations among the following mechanical
properties: E - Youngs modulus; Q - Poissons ratio; N - bulk modulus; G - shear modulus,
and the Lam constants ( M , N ).
Now, using the information in the table, we can show the elasticity tensor in terms of the
parameters ( E ; Q ) , (M; N G ) , ( N; N G ) , i.e.:
Ce
E
QE
1 1
I
(1 Q)(1 2Q)
(1 Q)
Ce
M1 1 2NI
Ce
N1 1 2N I 1 1
3
Elasticity tensor
(7.91)
(7.92)
1
{ De
Ce
1
{ De
Ce
1
{ De
(1 Q)
Q
1 1
I
E
E
1
M
1 1
I
2N(3M 2N)
2N
1
1
1
1 1
I 1 1
9N
2N
3
398
f (G; E )
f (G; N)
f (G; M)
f (G; Q)
f ( E; Q)
f ( N; M )
f ( N; Q )
7.5.3
G E 2G
3G E
2G
N
3
E 2G
2G
3N 2G
23N G
M G
2M G
E
31 2Q
2G
1 2Q
N9 N 3E
9N E
QE
1 Q 1 2Q
3NQ
1 Q
9GN
3N G
G 3M 2G
MG
2G 1 Q
3NE
9N E
E
21 Q
3N M
2
3N1 2Q
21 Q
f ( E; N)
N
GE
9G 3E
9 NN M
3N M
3N1 2Q
2
3
2G 1 Q
31 2Q
M
Q
3N E
6N
M
3N M
One significant tensor in elasticity is the elastic acoustic tensor ( Q e (N) ) defined as:
) N
Ce N
Q e (N
(7.93)
Q
e
jl
Ce N
N
i ijkl k
>
ME E N E E E E
N
i
ij kl
ik jl
il jk N k
ME E N
N
i
ij kl k NN i E ik E jl N k NN i E il E jk N k
MN j N l NN k N k E jl NN l N j
(7.94)
NE jl (M N) N j N l
which in tensorial notation becomes Q e (N ) N1 M N N N , and whose inverse form
is given by:
Qe
1
M N N N
1
1
M 2N
Q ejl
1
M N N N
1
E
N jl M 2N j l
(7.95)
1
2(1 Q )
1
N
1
N
21 Q
E
(7.96)
Next, the isotropic elastic acoustic tensor determinant can be evaluated as follows:
Qe
N 2 M 2N
(7.97)
NM 2N .
7 LINEAR ELASTICITY
399
Then to obtain the eigenvalues of Q e one need only solve the following characteristic
determinant:
>M N N1N1 N@ 9
M N N1N 2
M N N1N3
M N N1N 2
>M N N2N2 N@ 9
M N N 2N3
M N N1N3
M N N 2N3
>M N N3N3 N@ 9
(7.98)
which gives rise to the characteristic equation. Then, if we use the constraint
N12 N 22 N32 1 , we can obtain the following eigenvalues:
(Q e ) cij
N 0
0 N
0 0
M 2N
0
0
(7.99)
The system for small perturbation is unstable in the presence of zero or negative roots. In
other words, Q e must be a positive definite tensor in order to guarantee the stability of the
system. Now, the necessary and sufficient conditions for strong ellipticity occurs when N ! 0
and M 2N ! 0 , but if said strong ellipticity conditions are violated, the material is
subjected to instability shown by a homogeneous deformation band. These conditions can
also be expressed as follows:
E ! 0
E
Q ! 1
!0
2(1 Q)
E 0
Q 1
and
M 2N 2N
Q 0.5
(1 Q)
!0
(1 2Q)
Q ! 1
(7.100)
E 0 Q @ f ; 1>
and
(7.101)
Now, in order to have physical meaning the bulk modulus ( N ) (Truesdell&Noll 1965)
must be positive and the stability condition point-to-point is ensured by:
N!0
N M 23 N
E
!0
3(1 2Q)
(7.102)
Then for isotropic linear elastic material, strain energy is positive when it holds that:
E !0
1 Q 0 .5
(7.103)
For most materials, Poissons ratio is between the range 0 Q 0.5 and materials with
negative Poissons ratio are called auxetic materials.
400
In order to physically interpret the strain energy we will consider a differential volume
element dxdydz in which we have the normal stress V x , (see Figure 7.8).
y
Vx
dx
dz
Vx
dy
Vx
Stored energy
Vx
Hx
a)
Hx
b)
0 Vx
dydz (H x dx)
2
1
V x H x dxdydz
2
(7.104)
Likewise, we can obtain the strain energy caused by the normal stress types V y and V z .
Now, if we consider the shear stress W xy , (see Figure 7.9), the strain energy is given as
follows:
Moment u Angle
U0
U0
0 W xy
dxdz u dy u J xy
(7.105)
1
W xy J xy dxdydz
2
(7.106)
Next, if we consider the 6 Cauchy stress tensor components, the strain energy stored in a
differential volume element is:
U0
1
V x H x V y H y V z H z W xy J xy W xz J xz W yz J yz dxdydz
2
(7.107)
We can now introduce the strain energy per unit volume, : e , which is known as the strain energy
density and is given by:
:e
1
V x H x V y H y V z H z W xy J xy W xz J xz W yz J yz
2
J
m3
(7.108)
Then, the total strain energy ( U ) in the entire continuum can be evaluated by integrating
the strain energy density over the volume, i.e.:
U
dV
>Nm { J @
(7.109)
7 LINEAR ELASTICITY
401
y
W xy
dx
dz
J xy
W xy
dy
dy
x
dx
Indicial notation
1
:e
V ij H ij
2
1
:
2
:e
(7.110)
Then, using the generalized Hookes law C e : , the elastic potential becomes:
Tensorial notation
Indicial notation
1
:e
H ij H kl C ijkl
2
1
: Ce :
2
:e
Voigt notation
1
^ `T >C @^ `
:e
2
(7.111)
Next, if we consider the equations in (7.87) and substitute them into equation (7.110), we
obtain:
:e
:e
E
2
M
2
Q
Hx Hy Hz
(
1
)(
1 2Q )
Q
>H
Hy Hz
1
1
H 2x H 2y H 2z
J 2xy J 2yz J 2zx
1 Q
2(1 Q)
(7.112)
Additionally, if we take the derivative of the above equation with respect to the strain H x ,
we can obtain:
E
2
Q
Hx
2 Hx Hy Hz
1 Q
2 (1 Q)(1 2Q)
w: e
wH x
Vx
(7.113)
Vx ;
w: e
wH y
Vy ;
w: e
wH z
Vz ;
w: e
wJ xy
W xy ;
w: e
wJ yz
W yz ;
w: e
wJ zx
W zx
(7.114)
Indicial notation
w: e
w ij
ij
(7.115)
402
w 2: e
w w
7.6.1
w2 1
e
e
:C : C
w w 2
(7.116)
Next, we can split the strain energy density into deviatoric and spherical parts. To do so, let
us consider the strain energy density, (see Eq. (7.110)), for an isotropic linear elastic
material:
1
:
2
:e
MTr ( )
1
:
: >MTr ( )1 2N @
1 N :
2
2
Tr ( )
M>Tr ( )@2
M>Tr ( )@2
N Tr ( T )
N Tr ( )
M>Tr ( )@2
2
M>Tr ( )@2
2
N :
(7.117)
N Tr ( 2 )
where we have taken into account the constitutive equation in (7.66). The strain tensor can
be split into spherical and deviatoric parts, i.e. dev
Tr ( )
1 , which can be substituted
3
:e
M>Tr ( )@2
2
M>Tr ( )@2
2
N :
M>Tr ( )@2
2
Tr ( ) dev Tr ( )
N dev
1 :
1
3
3
>Tr()@2 1 : 1
Tr ( ) dev
Tr ( )
N dev : dev
:1
:
1
dev
,
3
3
9
3
Tr ( dev ) 0
Tr ( dev ) 0
N dev : dev N
>Tr()@2
3
2N >Tr ( )@
N dev : dev
M
3
2
2
N>Tr ( )@
N dev : dev
2
2
(7.118)
where:
:e
N
>Tr( )@2 N dev :
dev
2
purely volumetric
energy
purely distortional
energy
(7.119)
Then, we can conclude that the strain energy density allows for additive decomposition
into purely volumetric and distortional parts.
Next, if we consider that instead of substituting the equation of stress in (7.117) we
substitute the strain equation given in (7.69), the strain energy density becomes:
1
:
2
:e
M
1
1
:
Tr ( )1
N
M
N
N
2
2
2
(
3
2
)
1
M
Tr ( )
:1
:
,
4
N
4N(3M 2N )
Tr ( )
(7.120)
The double scalar product : is a scalar and an invariant. Now, using the stress principal
space
we
can
obtain
: V ij V ij V12 V 22 V 32 .
Then,
given
that
Tr ( )
7 LINEAR ELASTICITY
I 2
2( V 1 V 2 V 1 V 3 V 2 V 3 )
II
403
(V 1 V 2 V 3 ) 2
V12
V 22
V 32
V12 V 22 V 32
I 2 2 II
(7.121)
Now, if we also consider the equation of the second invariant of the deviatoric stress
tensor II dev
I 2
I2
II II dev (see Chapter 1),
3
3
2
2
I
I
I 2 2 J 2
2 J 2 : , with
3 3
II
which the strain energy density (7.120) can also be expressed as:
:e
M
1 I 2
2 J 2
I 2
4N(3M 2N)
4N 3
:e
1
I 2
6(3M 2N)
purely volumetric
energy
1
J2
2N
purely distortional
energy
(7.122)
Problem 7.1: Given an isotropic linear elastic material whose elastic properties are
E 71 GPa , G 26.6 GPa , find the strain tensor components and the strain energy density
at the point in which the stress state, in Cartesian basis, is represented by:
V ij
20 4 5
4 0 10 MPa
5 10 15
thus:
E
1 0.335
2G
1
1
>V11 Q V 22 V 33 @
>20 0.335 0 15@10 6
E
71 u 10 9
1
1
>V 22 Q V11 V 33 @
>0 0.335 20 15@10 6
E
71 u 10 9
1
1
>V 33 Q V11 V 22 @
>15 0.335 20 0 @10 6
E
71 u 10 9
1 Q
1 0.335
( 4 u 10 6 ) 75 u 10 6
V 12
E
71 u 10 9
1 Q
1 0.335
(5 u 10 6 ) 94 u 10 6
V 13
E
71 u 10 9
1 Q
1 0.335
V 23
(10 u 10 6 ) 188 u 10 6
E
71 u 10 9
H ij
E
2(1 Q )
211 u 10 6
165 u 10 6
117 u 10 6
211 75 94
75 165 188 u 10 6
188 117
94
Then, the strain energy density for an elastic material is obtained by the equation:
:e
1
: Ce :
2
1
:
2
indicial
o
:e
1
H ij V ij
2
Next, by considering the symmetry of the tensors and , the strain energy density can
be calculated as follows:
404
:e
1
>H11 V11 H 22 V 22 H 33 V 33 2H12 V12 2H 23 V 23 2H13 V13 @
2
1
>(211)(20) (165)(0) (117 )(15) 2( 75)(4) 2(188)(10) 2(94)(5)@ 5637 .5 J / m 3
2
We can also obtain the strain energy density by using the equation in (7.122), i.e.:
:e
1
I 2
6(3M 2N)
1
II dev
2N
1
I 2
6(3M 2N)
1
J2
2N
1 N :
2
M >Tr ( ) @
M >Tr ( ) @
N :
2
M >Tr ( ) @2
N Tr ( )
2
Tr ( )
N Tr ( T )
M>Tr ( )@2
2
N Tr ( 2 )
We can add and subtract the term N>Tr ( )@2 without altering the above outcome:
M>Tr ( )@2
1
2
2
M 2N >Tr ( )@2 N >Tr ( )@2 Tr ( 2 )
:e
N>Tr ( )@ N Tr ( 2 ) N>Tr ( )@
2
Finally, if we consider that the principal invariants of the strain tensor are I
II
1
>Tr ( )@2 Tr ( 2 ) , we can obtain:
2
Tr ( ) ,
1
M 2N I 2 2N II
2
Law
for
Orthotropic
For orthotropic material the stress-strain relationship is given by the following equation:
Hx
H
y
Hz
J xy
J yz
J zx
1
E
1
Q 12
E1
Q
13
E1
Q 21
E2
1
E2
Q 23
E2
Q 31
E3
Q 32
E3
1
E3
1
G12
1
G 23
0 V x
V y
0
Vz
W
0 xy
W yz
0 W zx
1
G13
(7.123)
7 LINEAR ELASTICITY
405
Q 12
E1
Q 31
E3
Q 13
E1
Q 32
E3
Q 23
E2
(7.124)
Next, the reciprocal of (7.123) provides the generalized Hookes law for orthotropic
material:
V x
V
y
V z
W xy
W yz
W zx
E1 (Q 32 Q 23 1)
E1 (Q 21 Q 23 Q 31 )
D
E (Q Q Q )
1
31
32 21
E1 (Q 21 Q 23 Q 31 )
E1 (Q 31 Q 32 Q 21 )
E 2 (Q 13 Q 31 1)
E 2 (Q 32 Q 12 Q 31 )
E 2 (Q 32 Q 12 Q 31 )
E 3 (Q 21Q 12 1)
G12
G 23
0
Hx
0 Hy
Hz
0 J
xy
0 J yz
0 J zx
G13
(7.125)
where: D Q 32 Q 23 Q 31Q 13 Q 21Q 12 2Q 21Q 13 Q 32 1 .
Note that when E1 E 2 E 3 ; Q 12 Q 13 Q 23 Q 21 Q 31 Q 23 ; G12 G 23 G13 are
satisfied, we revert to the isotropic case and thereby obtain the equations in (7.85) and
(7.86).
>C @
M
M0
M 2N
M
M
N
M0
2
M0
M0
M 0 2N 0
0
0
0
0
0
0
0
0
0
0
0
0
0
N0
0
0
0
0
0
N 0
(7.126)
Now, by decoupling the elasticity matrix into an isotropic and anisotropic part we obtain:
>C @
'M
M
M
0
0 0 0 0
M 2N
M
0
'M
M
N
M
2
0
0
0
0
N
0
0
0
0
0
0
0
0
0
0
0
0 N 0 0
0
0
0
0
0 0 N 0
0
0
0
0
0
0
0 0
0
0 'N 0
0 0 'N
0
0
0
0
0
(7.127)
where we have taken into account that 'M M M 0 and 'N N N 0 . Moreover, the
strain energy density, : e , can also be split into an isotropic and anisotropic part:
406
:e
e
: iso
: eAni
(7.128)
e
The isotropic part of the strain energy density, : iso
, is the same as that seen previously for
isotropic materials. The stress-strain relationship for the anisotropic part is considered as
follows:
V11
V
22
V 33
V12
V 23
V13
'M
0
0
0
'M
0
0
0
0
0
0
0
0
0
0
0 H11
0 H 22
0 H 33
0 0
0 2H12
0 'N 0 2H 23
0 0 'N 2H 13
0
0
0
0
0
(7.129)
Then, the anisotropic part of the strain energy density is given by:
: eAni
1
V ij H ij
2
1
V11H11 V 22 H 22 V 33 H 33 2V12 H12 2V 23 H 23 2V13 H13
2
(7.130)
1
^'MH 33 H11 'MH 33 H 22 >'MH11 'MH 22 'M 2'N H 33 @ H 33
2
0 4'NV 23 H 23 4'NV13 H 13 `
(7.131)
'M 2
2
2
'M>Tr ( )@ H 33 'N
H 33 2'N H13 H 23
2
(7.132)
FORCE
SYSTEM (I)
FORCE
SYSTEM (II)
F
A
Disturbed zone
Undisturbed zone
Disturbed zone
F
A
7 LINEAR ELASTICITY
407
The superposition principle states that the balance of a system in which several actions take
place is equal to the sum of all independent actions, (see Figure 7.11). This principle is valid
because the governing equations have been linearized. As example, we can decouple the
thermo-mechanical process, i.e. we can treat the different parts independently.
=
&
u
+
&
u( )
&
(u)
( )
&
u( (T ))
(T )
( (T ))
1m
100 m
1m
y, v
Vy
x, u
z, w
6000
1u1
6000 N
Figure 7.12
Solution: Using the normal strain expressions we can obtain:
Hx
Hy
Hz
>
1
Vx Q Vy Vz
E
>
1
V y Q V x V z
E
1
Vz Q Vx Vy
E
>
Q
Vy
E
(0.3)(6000 )
10 7
0.00018
Vy
6000
0.0006
E
10 7
Q
V y 0.00018
E
408
Initial volume
Final volume
&
H ij (u) H ij ('T )
(7.133)
&
&
where H ij (u) shows the mechanical strain in terms of the displacement field ( u ) and
H ij ('T ) is the thermal strain in terms of the temperature variation ( 'T ).
(7.134)
where T0 is the initial temperature; T is the final temperature, and B is the coefficient of
thermal expansion. For further details about thermomechanical problems, see Chapter 10.
7 LINEAR ELASTICITY
409
Next, we will show the coefficients of thermal expansion for some materials, namely:
B steel 12 u 10 6
1
, B aluminum
C
23 u 10 6
1
, B copper
C
17 u 10 6
1
.
C
M
1
V E
V B (T T0 )E ij
N(3M 2N) kk ij 2N ij
Q
1 Q
V kk E ij
V ij B(T T0 )E ij (7.135)
E
E
The Hookes law for isotropic materials including the thermal effect is given by the
reciprocal of the equation in (7.135), the result of which is:
V ij
V ij
QE
E
E
H kk E ij
H ij
B (T T0 )E ij
(1 Q)(1 2Q )
(1 Q)
(1 2Q)
(7.136)
Problem 7.4: Let us consider a length rod equal to L 7.5m , whose diameter is equal to
0.1m , which is made up of a material whose properties are: E 2.0 u 10 11 Pa and
B 20 u 10 6
1
. Initially the rod has a temperature equal to 15 C which later rises to
C
50 C .
1) Considering that the rod can expand freely, calculate the total elongation of the rod, 'L ;
2) Now assume that the rod can not expand freely because concrete blocks have been
placed at its ends, (see Figure 7.14(b)). Find the stress in the rod.
Hint: Consider the problem in one dimension.
x
'L
'L(1) 'L( 2 )
'L(1)
'T
'T
L
L
'L( 2)
b)
a)
Figure 7.14: Rod under thermal effect.
Solution: 1) To obtain the elongation, we pre-calculate the thermal strain according to the
rod axis direction H ij B ('T )E ij . Since this is a one-dimensional case, we need only
consider the normal strain component according to the x -direction, H 11 H x , then:
H 11
20 u 10 6 (50 15) 7 u 10 4
Hx
(1)
Then, the total elongation, 'L 'L 'L( 2) , is obtained by solving the integral:
L
'L
x dx
HxL
7 u 10 4 u 7.5 5.25 u 10 3 m
410
MH kk E ij 2NH ij
V ij , j
MH kk , j E ij 2NH ij , j
(7.137)
Furthermore, by using the kinematic equations in (7.6) we can obtain the term H ij, j and
therefore H kk , j , i.e.:
1
u i , j u j ,i
2
H ij
2H ij , j
i , jj
u j ,ij
H kk , j
u k ,kj
(7.138)
Mu k ,kj E ij Nu i , jj u j ,ij (M N) u j , ji Nu i , jj
(7.139)
Finally, by substituting the equation (7.139) into the equations of motion given in (7.4),
V ij , j Sb i Su i , we can obtain:
(M N)u j , ji Nu i , jj Sb i
i
Su
& &
&
&
&
w 2 u( x , t )
(M N) u N 2 u Sb S
wt 2
(7.140)
which are known as the Navier-Lam equations. With that we have reduced the number of
equations as well as the number of unknowns. Note that the only remaining unknowns are
the displacement components. Finally, as for addressing specific problems, this equation
can be used to obtain an analytical solution of the linear elasticity problem.
It should be stressed that such simplification are mere approximations of the real
problem. Nevertheless, in many cases they turn out to be quite satisfactory, i.e. the
error made when using them are insignificant.
2) Simplification on a mathematical level
We use these simplifications in structures that have radial symmetry. Such structures
are known as:
7 LINEAR ELASTICITY
411
The results obtained by using this simplification are exactly the same as considering
the problem from a three-dimensional point of view.
x2 , y
x1 , x
x3 , z
V11
V
12
0
V12
V 22
0
0
0
0
V x
W
xy
0
W xy
Vy
0
0
0
0
(7.141)
412
Hx
Hy
Hz
J xy
J yz
J zx
1
E
Q
E
Q
E
Q
E
1
E
Q
E
Q
E
Q
E
1
E
1
G
1
G
Vx
Vy
0
Vz
W xy
0
W yz 0
W zx
(7.142)
Then, if we remove the columns and rows associated with the zero stresses, the stressstrain relationship for the plane stress case is given by:
Hx
Hy
J xy
1
E
Q
E
0
Q
E
1
E
0
0
Hx
V x G E
(1 Q )
o H y
0 V y 2
J xy
1 W xy
0 V x
1 Q
1
Q
1
0 V y
E
0
0 2(1 Q ) W xy
(7.143)
The reciprocal of the above equation will result in Hookes law for the state of plane stress:
1 Q
0 Hx
E
Q 1
0 Hy
1 Q2
1 Q
J xy
0 0
2
V x
V y
W xy
>
1
Vz Q Vx Vy
E
Hz
>
1
Q Vx Vy
E
(7.144)
Note that the normal strain H z is not equal to zero, since H z is not just dependant on the
normal stress V z . Then, the strain tensor components are represented as follows:
H ij
Hx
1
2 J xy
0
1
2
J xy
Hy
0
0
H z
(7.145)
Ce
1
: 0
Ce : Ce : 0
(7.146)
If we are considering thermal effects, the equations for the state of plane stress become:
7 LINEAR ELASTICITY
Strain:
Hx
Hy
J xy
1
E
Q
E
0
Q
E
1
E
0
0
1
V x
0 V y B'T 1
0
1 W xy
(7.147)
Q
V x V y B 'T
E
Hz
413
Stress:
1 Q
0 Hx
E
EB'T
Q 1
0 Hy
1 Q2
1 Q 1 Q
0 0
J xy
2
V x
V y
W xy
1
1
0
(7.148)
2D
y
x
p
x
Cross section
per unit length
prismatic axis
Figure 7.17: Cylinder under pressure.
414
2D
1
W xy
W yz
W zx
we obtain:
Q
Q
0
1 Q
Q
Q
Q
1
0
Q
Q 1 Q
0
1 2Q
E
0
0
0
2
(1 Q)(1 2Q )
0
0
0
0
0
0
0
0
0
0
0
0
1 2Q
2
0
Hx
H
y
Hz
0 J xy
0 J yz
1 2Q J
zx
2
0
0
0
(7.149)
7 LINEAR ELASTICITY
415
Q
0 Hx
1 Q
E
Q 1 Q
0 Hy
(1 Q )(1 2Q)
1 2Q
0
0
J xy
2
V x
V y
W xy
(7.150)
Vz
(7.151)
1 Q Q 0 V x
1 Q
Q 1 Q 0 V y
E
0
2 W xy
0
(7.152)
Afterwards, we can write the constitutive law for the state of stress and strain in a single
equation as:
V x
V y
W xy
1
E
Q
1 Q2
0
0 Hx
1
0 Hy
1 Q
0
J xy
2
(7.153)
E ; Q
E
1 Q2
; Q
Q
1 Q
(7.154)
0 Hx
Q
1 Q
E
EB'T
Q 1 Q
0 Hy
(1 Q)(1 2Q)
1 2Q 1 2Q
0
0
J xy
2
1
1
0
(7.155)
Note that the above equation is the same as that given in (7.136) when we are dealing with
two-dimensional cases, i.e. when i, j 1,2 . Our goal now is to obtain the strain field and to
do so, we will restructure the above equation as:
V x
EB'T
V y 1 2Q
W xy
1
1
0
0 Hx
Q
1 Q
E
Q
1 Q
0 Hy
(1 Q)(1 2Q)
1 2Q
0
J xy
0
2
Then if we multiply the above equation by the matrix given in (7.152) we can obtain:
(7.156)
416
1 Q Q 0 V x
1 Q Q 0 1
1 Q
V EB'T 1 Q Q 1 Q 0 1
1
0
Q
Q
y 1 2Q E
E
0
0
2 W xy
0
2 0
0
Hx
Hy
J xy
(7.157)
1 Q Q 0 V x
1
1 Q
V (1 Q)B'T 1
Q
1
Q
0
y
E
0
0
0
2 W xy
(7.158)
Problem 7.5: A strain gauge (or strain gage) is a device used to obtain the strain in only one
direction. Consider a strain rosette that contains three strain gauges where there are 45
internal angles, (see Figure 7.20). At one point we have calculated the following strain
values:
Hx
0.33 u 10 3
Hcx
0.22 u 10 3
Hy
0.05 u 10 3
strain gauge
45
45
H 11 H 22 H 11 H 22
cos( 2T) H 12 sin( 2T)
2
2
The above equation was obtained by means of the transformation law, (see Chapter 1),
which in engineering notation becomes:
Hcx
Hx Hy
2
Hx Hy
2
cos( 2T)
J xy
2
sin( 2T)
7 LINEAR ELASTICITY
417
(H H y ) (H x H y )
2
Hcx x
cos( 2T)
sin( 2T)
2
2
J xy
0.16 u 10 3
thus
H ij
0.33 0.08 0
0.08 0.05 0 u 10 3
0
0
0
>
>
E
(1 2Q )H x QH y 12 .0462 Pa
(1 Q )(1 2Q )
E
(1 2Q )H y QH x 3.5692 Pa
(1 Q )(1 2Q )
E
EQ
J xy 1.7846 Pa ; V z
Hx Hy
2(1 Q )
(1 Q )(1 2Q )
>
4.684 Pa
Vx Vy
W 2xy
4.5988 Pa
(u 10 3 )
Then, by solving the above equation we can find the eigenvalues (principal strains) given
by:
H1
0.346155 u 10 3
0.06615528 u 10 3
H2
12.0462 1.7846
1.7846 3.5692
0
0
0
0 Pa
4.684
We now obtain the characteristic determinant and in turn the eigenvalues (principal
stresses) V 1 12.40654 , V 2 3.208843 . Additionally, the eigenvectors of the stress tensor
are:
V1 0.9802
V 2 0.1979
V 3
0
0.1979
0.9802
0
0
0
1
As expected, the eigenvectors of stress and strain are the same; since we are working with
isotropic linear elastic material.
wu
wr
Hz
ww
wz
J rz
wu ww
wz wr
(7.159)
418
where H r is the radial strain, H z is the axial strain, and J rz is the shear strain.
We can then introduce the strain in the circumferential direction H T as:
HT
2S(rP u ) 2SrP
2SrP
u
rP
(7.160)
wu
wr
ww
wz
wu ww
wz wr
(7.161)
Next, the generalized Hookes law for a solid of revolution is given by:
V r
V
T
V z
W rz
0 H
Q
Q
1 Q
r
Q
1 Q
0 H
Q
E
T
0 H
Q 1 Q
(1 Q)(1 2Q) Q
1 2Q z
0
0
0
J
2 rz
(7.162)
Vz
W rz
VT
Vr
7 LINEAR ELASTICITY
419
deformation remains planar. Consequentially, the strain and stress fields at the beam cross
section are defined by planes, (see Figure 7.22).
It must be pointed out that in the deep beam case, (see Figure 7.16), the approach adopted
in this subsection is invalidated, since the beam cross section does not remain planar after
deformation (bending).
Strain diagram
z
z
H x ( z)
y
y
Stress diagram
V x ( z)
neutral axis
a) beam
If we now make a cut in a cross section according to the orientation of the plane 3 , in
general, the stress state at a point in this cross section is given as shown in Figure 7.23. The
intensity (or even the nonexistence) of stress depends on the load type (external force) and
on the beam cross section.
W xz
W xy
Vx
x
A - cross section area
420
V x ( y, z )
x
V (x1)
V (x2 )
y
V
A
V (x3)
(1)
x dA
My
zV
A
(2)
x dA
Mz
yV
( 3)
x dA
Mz
My
( 2)
x dA
( 3)
x dA
Figure 7.24: The internal normal force and the bending moments.
Then, by considering Figure 7.25, the bending moment M y is defined as follows:
My
V
A
x zdA
VS z
zdA
c
A
VS
c
z
A
dA
VS
Iy
c
(7.163)
7 LINEAR ELASTICITY
421
where I y is the moment of inertia of the cross section about the y -axis. Then, if we
observe that
VS
c
Vx
, we can obtain:
z
V x ( z)
My
(7.164)
Mz
y
Iz
(7.165)
Iy
VS
c
Vx
neutral axis
xz y
W yz z dA
(7.166)
W xz ( y , z )
W xy ( y, z )
W
A
xz dA
xy dA
Qz
Qy
422
z
W xz
W xy
W max
W(r )
b) Circular cross-section
a) Rectangular cross-section
1 V (x1)
dV
2V E
1 (1) (1)
V x H x dV
2V
1 N2
dAdx
2 0 EA2 A
1 N2
dx
2 0 EA
Likewise, we can obtain the strain energy associated with the normal stress V(x2)
U
1 ( 2) ( 2)
V x H x dV
2V
1
20
My
Iy
My
EI y
zdAdx
L
2
1 My 2
z dAdx
2 0 EI y2 A
L
2
1 My
dx
2 0 EI y
(7.167)
EH (x2 ) as:
(7.168)
1 M z2
dx
2 0 EI z
(7.169)
Then, if we follow the same procedure for the other stress components, we can obtain the
strain energy of a bar in function of the internal forces as:
U
L
2
2
1 N 2 M y M z2 9Q y 9Qz2 M T2
dx
2 0 EA EI y EI z GA GA EJ T
(7.170)
where 9 is the correction factor for the cross-section, and J T is the effective polar
moment.
8 Hyperelasticity
Hyperelasticity
8.1 Introduction
Some materials such as elastomers, polymers, rubber and biological matter (arteries,
muscles, skin, etc.) may be subject to large deformations without there being any internal
energy dissipation (which is typical en elastic process). These materials are classified as
being hyperelastic and purely hyperelastic materials have no memory of motion history, i.e.
they are only dependent on the current values of the state variables.
Physically speaking, elastic materials (linear elasticity, hyperelasticity) return to their initial
state once their load disappears, (see Figure 8.1). In other words, the work done during the
loading process is recovered during the unloading process, i.e. there is no internal energy
dissipation (a reversible process).
Our goal in this chapter is to establish the constitutive equations for materials that behave
according to the hyperelasticity theory, also known as Green or nonlinear elasticity. Moreover, we
will limit our analysis to purely mechanical theories, so we have eliminated thermodynamic
variables such as temperature and entropy.
Among the researchers who have used the hyperelastic constitutive equations to model
rubberlike materials we can mention: Alexander (1968), Treloar (1975), Ogden (1984),
Morman (1986) and Holzapfel (2000).
423
424
II
Dint : D : 0
Dint
P : F : 0
1
S : C : 0
2
:
: D (Current configuration)
: P : F
(Reference configuration)
1
:
S : C
(8.1)
where is the Cauchy stress tensor, D is the rate-of-deformation tensor, P is the first
Piola-Kirchhoff stress tensor, S is the second Piola-Kirchhoff stress tensor, and C is the
right Cauchy-Green deformation tensor. Then, taking into account the conjugate relations
obtained in Chapter 5 we have:
S
J : D dV
: DdV ,
: DdV S : E dV 2 S : CdV P : F dV S
V0
V0
V0
V0
V0
P : F dV
(8.2)
Then, in summary we can state that the rate of change of the strain energy density can also
be expressed as follows:
1
S : C : D
2
: P : F S : E
(8.3)
Stress Power
where E is the Green-Lagrange strain tensor, and is the Kirchhoff stress tensor with
which we can state that a material is considered to be hyperelastic if and only if the rate of
change of the strain energy is equal to the stress power.
8 HYPERELASTICITY
425
wt
w: ( F )
:F
wF
(8.4)
Next, by substituting the above equation into the internal energy dissipation given by the
equation in (8.1) we obtain:
w: ( F )
P : F
:F
wF
P : F
w: ( F )
:F
wF
(8.5)
(8.7)
:
: (C )
: (E)
where U is the right stretch tensor and R is the polar decomposition rotation tensor, (see
Chapter 2). Since the tensors C , U , and E are directly linked by C U 2 , and 2 E C 1 ,
the strain energy density can also be expressed in terms of the Green-Lagrange strain
tensor ( E ). Then, similarly to (8.6), it is possible to express the constitutive equation for
stress in the material description. Next, if we take the rate of change of the energy : (C )
we can obtain:
w: (C ) wC
: (C )
:
wC
wt
w: (C )
:C
wC
(8.8)
Dint
1
S : C : (C ) 0
2
1
w: (C )
S : C
:C 0
2
wC
w: (C )
1
S
:C 0
wC
2
(8.9)
Note that, the condition in (8.9) must hold for any thermodynamic process. Now, in a
mechanical process with C z 0 , the only scenario in which the condition (8.9) remains
valid is when:
1
w: (C )
S
2
wC
w: (C )
wC
(8.10)
426
w: (C )
wC
w: ( E )
wE
R U
reference
configuration
(8.11)
current
configuration
F , : (F )
B0
&
X
C, E
QF
: (Q F )
F*
: (C )
: ( E )
&
x
Q RT
U
: (Q F ) : (Q R U) : (U)
: (U)
&
X
F S F T
2F
F S F T
2F
w: (C )
FT
wC
(8.12)
J P F T
w: (C )
FT
wC
w: ( F )
F T
wF
(8.13)
w: ( E )
FT
wE
F T
F S F T F T
F S
F 2
w: (C )
wC
(8.14)
C S
F T P
w: (C )
wC
w: ( F )
FT
wF
2C
(8.15)
8 HYPERELASTICITY
427
Hence, we can sum up the different ways of expressing the stress constitutive equations for
hyperelastic materials as:
P
w: ( F )
wF
F 2
w: (C )
;
wC
w: ( E )
wE
w: (C )
2
wC
2F
w: (C )
FT
wC
w: (C )
; M 2C
wC
w: ( F )
F
wF
T
Stress constitutive
equations for
hyperelastic materials
(8.16)
The normalization condition: : (1) 0 , i.e. the strain energy function vanishes
when the material has been completely unloaded, i.e. when F 1 ;
In a reversible process (without internal energy dissipation) the following must be satisfied:
P : F dt
t1
t2
S : E dt
t1
t2
2 S : C dt
(8.17)
t1
Now, for an elastic material (reversible) the internal work done is independent of the path,
so, the following must be met:
P : dF *c P : dF
(8.18)
For any closed cycle of deformation the work done is equal to zero:
P : dF
8.2.1
8.2.1.1
or
S : dE
(8.19)
The rate of change of the constitutive equation in (8.16), S (E ) , can be expressed as:
S
w 2: (E)
:E
wE wE
S ij
C tan : E
w 2: (E)
E kl
wE ij wE kl
C tan
ijkl E kl
(8.20)
where S , E are objective rates, and C tan is a fourth-order tensor known as the material
elastic tangent stiffness tensor also called the material tangent elasticity tensor. Remember that the
tensors E and C are related to each other by the equation 2 E C 1 2 E C , thus:
S
w 2 : (C )
:C
wC wC
S
w 2 : (C )
:E
4
C
C
w
w
(8.21)
C tan
Then, taking into account the equations in (8.11), (8.20) and (8.21) we can conclude that:
428
w 2: (E)
wE wE
C tan
wS
wE
w : (C )
4
wC wC
2
wS
2
wC
S ji , E ij
(8.22)
C tan
jikl
C tan
ijlk
C tan
jilk
(8.23)
Then, taking into account the equation in (8.22) we can conclude that the tensor C tan also
has major symmetry:
w 2:
wE ij wE kl
C tan
ijkl
w:
wE kl
w
wE ij
w
wE kl
w:
wE ij
C tan
klij
(8.24)
Therefore, we can conclude that the tensor C tan is symmetric. In the general case C tan is
anisotropic and has 21 independent components. For further details regarding symmetry
types see Chapter 7.
8.2.1.2
The rate of change of the second Piola-Kirchhoff stress tensor can be obtained by means
of the equation S F 1 F T , i.e.:
F 1 F T F 1 F T F 1 F T
S
F 1 l F T F 1 F T F 1 l
F
1
l l
F T
(8.25)
T
l l
(8.26)
Remember also that is not objective, but is. Then, by substituting (8.26) into (8.25) we
obtain:
S
(8.27)
F 1 F T
Then, given the relationship between the rate of change of the Green-Lagrange strain
tensor and the rate-of-deformation tensor we have:
E
F T D F
(8.28)
Additionally, by substituting the equations (8.27) and (8.28) into the constitutive equation
(8.20) we obtain:
S
F 1 F T
S mn
C tan : E
C tan : F T D F
Fms1 W st Fnt1
C tan
mnpq E pq
C tan
mnpq Fkp D kl Flq
(8.29)
Then, taking into account the symmetry of the Oldroyd rate of the Kirchhoff stress tensor
8 HYPERELASTICITY
F F
F F
1
429
T
F C
tan
:F
D F F
F jn Fim C tan
mnpq Fkp Flq D kl
E is W st E jt
F jn Fim C tan
mnpq Fkp Flq D kl
F F : C
tan
: F F
W ij
: D
F Fim C tan
mnpq Fkp Flq D kl
jn
Lijkl
(8.30)
(8.31)
L :D
where we have introduced the spatial elastic tangent stiffness tensor, also called the spatial
tangent elasticity tensor, in the current configuration, which is given by:
( F F ) : C tan : ( F T F T )
L ijkl
Fim F jn C tan
mnpq Fkp Flq
(8.32)
Then, from the above equation we can obtain the inverse relationship:
Fai1 Fbj1 L ijkl Fck1 Fdl1
1
1
Fai1 Fim Fbj1 F jn C tan
mnpq Fck Fkp Fdl Flq
E am E bn C tan
mnpq E cp E dq
(8.33)
C tan
abcd
Thus
Fai1 Fbj1 L ijkl Fck1 Fdl1
C tan
abcd
C tan
1
F 1 : L : F T F T
(8.34)
$
In Chapter 4 we obtained the relationship between the Jaumann-Zaremba rate ( ) and the
D D
D D
(8.35)
Next, by combining the above rate with the constitutive equation in (8.31) we can
obtain:
$
D D L : D
$
W ij
L :D D D
L ijkl D kl D ip W pj W ip D pj
(8.36)
Notice that the tensor D is symmetric, so the double dot product between the symmetric
fourth-order unit tensor, I sym , and a symmetric second-order tensor turns out to be the
same tensor, so,
sym
I ipkl
D kl
D ip
1
(E ik E pl E il E pk )D kl
2
D pj
I sym
pjkl D kl
1
(E pk E jl E pl E jk )D kl
2
(8.37)
W ij
1
1
(E ik E pl E il E pk ) W pj D kl W ip (E pk E jl E pl E jk )D kl
2
2
1
1
(W pj E ik E pl W pj E il E pk ) (W ip E pk E jl W ip E pl E jk ) D kl
2
2
L ijkl D kl
L ijkl
1
1
L ijkl 2 (W lj E ik W kj E il ) 2 (W ik E jl W il E jk ) D kl
>
(8.38)
430
(8.39)
L : D
where L is a fourth-order tensor and is defined by:
L
(8.40)
L 2H
with
1
(W lj E ik W kj E il W ik E jl W il E jk )
4
H ijkl
(8.41)
We can now summarize the relationships between the rate of change of the stress and the
rate-of-deformation tensor as:
L : D
W ij L ijkl D kl
$
L : D
$
W ij L ijkl 2H ijkl D kl
>
where
L ijkl
F jn Fim C tan
mnpq Fkp Flq
L ijkl
L ijkl 2H ijkl
H ijkl
8.2.1.3
(8.42)
The spatial elastic tangent stiffness tensor
(Current configuration)
1
(W lj E ik W kj E il W ik E jl W il E jk )
4
The relationship between the Cauchy stress tensor and the second Piola-Kirchhoff stress
tensors is given by S JF 1 F T whose rate of change becomes:
S
JF 1 F T JF 1 F T JF 1 F T JF 1 F T
(8.43)
1
JF 1 l l
S
T
Tr (D)
T
(8.44)
Now, remember in Chapter 4 that the Truesdell stress rate, which is objective, is given by
7
l l
(8.45)
JF 1 F T
Then, by substituting the equation (8.45) into the constitutive equation in (8.20) and if we
know that E F T D F , we can obtain:
S
7
JF 1 F T
7
or
C tan : E
C tan : F T D F
1
F F : C tan : F T F T : D
J
(8.46)
8 HYPERELASTICITY
7
431
(8.47)
A :D
where A is the instantaneous elastic tangent stiffness tensor, also called the instantaneous elastic
moduli, (see Asaro&Lubarda(2006)), which is defined by:
A
1
F F : C tan : F T F T
J
1
L
J
1
Fim F jn C tan
mnpq Fkp Flq
J
A ijkl
1
L ijkl
J
(8.48)
A :D
V ij
A ijkl D kl
with
A ijkl
(8.49)
1
Fim F jn C tan
mnpq Fkp Flq
J
1
L ijkl
J
Then, by taking into account the relationship between the Truesdell stress rate ( ) and the
Oldroyd rate of the Kirchhoff stress tensor , i.e. J , the equation in (8.47) becomes:
7
A :D
1
L :D
J
L :D
(8.50)
The constitutive equation for stress that relates the first Piola-Kirchhoff stress tensor to the
deformation gradient is:
P
w: ( F )
wF
w: (F )
wFij
Pij
(8.51)
Remember that F and P are two-point tensors (pseudo-tensors), i.e. they are not defined
in any configuration. Then, the rate of change of the above constitutive equation is given
by:
P
P
w 2 : (F )
:F
wF wF
P ij
K : F
P ij
w
wFij
w: (F )
Fkl
wFkl
(8.52)
K ijkl Fkl
We can now introduce the elastic tangent stiffness pseudo-tensor also called the elastic
pseudomoduli, (see Lubarda&Benson (2001)), as follows:
K
w 2 : (F )
wF wF
K ijkl
w 2 : (F )
wFij wFkl
w 2 : (F )
wFkl wFij
K klij
(8.53)
The elastic tangent stiffness pseudo-tensor is not a real moduli, because it is partially
associated with the material spin tensor, (see Asaro&Lubarda (2006)).
Next, we can relate the tensors K and C tan . To do so, we need to evaluate the rate of
change of Pij Fip S pj , (see Eq. (8.14)):
P ij
S pj Fip Fip S pj
(8.54)
432
Then, by substituting (8.21) and (8.52) into the above equation we obtain:
K ijkl Fkl
(8.55)
1
Fqk Fql Fqk Fql the above equation becomes:
2
1 tan
C pjkl Fqk Fql Fqk Fql Fip Fip S pj
2
1 tan
C pjkl Fql Fip Fqk C tan
pjkl Fqk Fip Fql Fip S pj
2
K ijkl Fkl
(8.56)
Note that the dummy indices k and l from the expression C tan
pjkl Fql Fip Fqk can be
exchanged without altering the result of the expression, and the dummy indices k and q
from C tan
pjkl Fqk Fip Fql can also be exchanged, so:
1 tan
C pjlk Fqk Fip Fql C tan
pjql Fkq Fip Fkl S pj Fip
2
1 tan
C pjlq Fkq Fip Fkl C tan
pjql Fkq Fip Fkl S pj Fip
2
K ijkl Fkl
S lj Fil
tan
pjlq Fkq Fip
tan
pjlq Fkq Fip
F
kl
(8.57)
C tan
pjql , we can still state that:
Fip S pj
(8.58)
S lj E ik Fkl
K ijkl
C tan
pjlq Fkq Fip S lj E ik
(8.59)
Thus:
P
K : F
K ijkl
w 2 : (F )
wFij wFkl
K ijkl
C tan
pjlq Fkq Fip S lj E ik
P ij
K ijkl Fkl
(8.60)
The elastic tangent stiffness pseudo-tensor
(8.61)
for any orthogonal tensor Q . Then if we use the polar decomposition rotation tensor, i.e.
Q R , and if we know that C F T F , we can obtain:
: (C ) : (R C R T ) : (R F T F R T ) : ( V T V ) : ( V 2 )
: (b)
(8.62)
8 HYPERELASTICITY
433
where V V T is the left stretch tensor which is related to the left Cauchy-Green
deformation tensor ( b F F T ) by means of b V 2 , where F V R V F R T is
satisfied, (see Chapter 2). Thus, in isotropic materials, the energy function : can be
expressed in terms of the left Cauchy-Green deformation tensor as follows:
&
&
: (C , X ) : (b, x )
Reference
configuration
P ij
(8.63)
Current
configuration
K ijkl Fkl
w 2 : (F )
wFij wFkl
K ijkl
C tan
pjlq Fkq Fip S lj E ik
F
B0
&
X
&
x
C tan : E
W ij
L ijklD kl Wij
D V
L
ijkl
kl
ij
A ijklD kl
tan
w 2 : ( E ) wS
wE wE wE
w 2 : (C )
wS
4
2
wC wC
wC
F jn Fim C tan
mnpq Fkp Flq
L ijkl
L ijkl 2H ijkl
with
H ijkl
1
(W lj E ik W kj E il W ik E jl W il E jk )
4
1
Fim F jn C tan
mnpq Fkp Flq
J
1
L ijkl
J
434
8.3.1
8.3.1.1
: (C ) : ( I C , II C , III C ) : ( I b , II b , III b )
Tr(C ) ,
I b , II C
1
>Tr(C )@2 Tr (C 2 ) , III C
2
II C
II b and III C
det(C )
(8.64)
III b , which is obtained
J2
1
3
1
3
3
2
Tr (C ) Tr (C ) Tr (C ) >Tr (C )@
3
2
2
w: wI C
w: w II C
w: w III C
wI C wC w II C wC
w III C wC
w:
w:
w:
I C 1
w
w
I
I
I
C
w II C
C
w:
wI C
w:
wI C
w:
C
III C
w III C
1
C
w:
w:
w:
1
I C C
w
w
I
I
I
I
I
C
w III C
w:
w:
w:
1
II C
III C C 1
III C C 2
w III C
w II C
w II C
w:
w:
IC
II C
w II C
w III C
2
C
(8.65)
IC 1 C T
IC 1 C
II C C 1 III C C 2 ,
(8.66)
III C C 1
C 2 I C C II C 1
Now, taking into account the equations in (8.11) and (8.65) we can obtain the constitutive
equation in terms of the principal invariants of C as follows:
S
w: (C )
wC
w:
w:
w:
2
I C 1
w II C
wI C w II C
w:
C
III C C 1
w III C
w:
w:
w:
w:
w:
w:
2
IC
II C 1
I C C
w
w
w
w
w
I
I
I
I
I
I
I
I
I
I
I
C
C
C
w III C
C
w:
2
wI C
2
C
(8.67)
w:
w:
w:
1
II C
III C C 1
III C C 2
w III C
w II C
w II C
In Chapter 1, in the subsection about the Tensor-Valued Tensor Function, it was shown
that the following relationships are valid:
8 HYPERELASTICITY
: , b b F : ,C F T
435
b : ,b
(8.68)
Now, if we consider the relationship between the Kirchhoff stress and the second PiolaKirchhoff stress tensors, F S F T , and the constitutive equation in the reference
configuration (8.11), S 2: ,C , it is possible to obtain the constitutive equation in the
current configuration as follows:
F S F T
F 2: , C F T
2: , b b
2b <,b
(8.69)
Next, by taking into account the equation J , we can also represent the constitutive
equation for isotropic materials as:
w: (b)
b 2b w: (b)
wb
wb
and
J 1 2
w: (b)
b
wb
J 1 2b
w: (b)
wb
(8.70)
w:
w:
w:
2 F
I C 1
C
III C C 1 F T
w
I
I
I
I
I
w
wI C w II C
C
C
J F S F T
F T F and b
w:
w:
(8.71)
w:
w:
J 2
I b b
b2
III b 1
w III b
w II b
wI b w II b
(8.72)
1
1
b I b b II b 1 III b b
b
I b b II b 1 III b b
(8.73)
w:
w:
w:
b
J 2
III b
II b 1
III b b 1
w II b
w Ib w IIb
w III b
(8.74)
F T
F S F T F T
F S
(8.75)
the constitutive equation can also be written in terms of the first Piola-Kirchhoff stress
tensor. To do so, let us consider S given in (8.67), and so we obtain:
P
w:
w:
I C 1
2F
wI C w II C
w:
w:
2
I C F 1
wI C w II C
w:
w:
C
III C C 1
w II C
w III C
w:
1
F
F
F
F T
FT F w: III C
w II C
w III C
b
1
(8.76)
w:
w:
w:
w:
I C 1
b
III C b 1 F
2
w II C
w III C
wI C w II C
w: (b)
F
wb
(8.77)
436
8.3.1.2
Energy can also be written in terms of the Green-Lagrange strain tensor E , and if we are
dealing with isotropic material the energy constitutive equation can be expressed in terms
of the principal invariants of E :
: ( E ) : ( I E , II E , III E )
1
>Tr ( E )@2 Tr ( E 2 ) , III E det(E ) . Then, if we consider the
2
w: ( E )
, we can obtain another one analogous to that obtained in
wE
Tr(E ) , II E
where I E
(8.78)
c0 1 c1 E c2 E 2
(8.79)
w:
w:
w:
IE
II E
wI E w II E
w III E
8.3.2
w:
w:
IE
w II E w III E
c1
w:
w III E
c2
(8.80)
c I
: : ( I C , II C , III C )
pqr
3
II C
3
III C
1
(8.81)
p ,q ,r 0
(a) N
(a)
N
U2
a 1
2
a
(a) N
(a)
N
(8.82)
a 1
Next, the principal invariants of C or b in terms of the principal stretches O i are given by:
C ij
U ij2
O21
0
0
0
O22
0
0
O23
2 2
2 2
2 2
II C II b O 1O 2 O 2 O 3 O 1O 3
2 2 2
III C III b O 1O 2 O 3
(8.83)
Then, by substituting the values of (8.83) into the power series (8.81) and after some
mathematical manipulations we obtain:
: : (O 1 , O 2 , O 3 )
a ^>O O
f
pqr
p ,q ,r 0
p
1
q
2
6
(8.84)
8 HYPERELASTICITY
437
c I
: : ( I C , II C )
pq
3
II C
3
(8.85)
p ,q 0
: : (O 1 , O 2 , O 3 )
p ,q 0
8.3.3
pq
^ >O O
p
1
q
2
@ `
(8.86)
As we have seen before, for isotropic materials, we can express the strain energy function
in terms of the principal stretches O a , a 1,2,3 , i.e. : : (O 1 , O 2 , O 3 ) . Let us now suppose
( a ) and n ( a ) are the principal directions (eigenvectors) of the right stretch tensor ( U )
that N
and the left stretch tensor ( V ), respectively, where the following holds:
U
(a) N
(a )
N
; E
a 1
2 (O
2
a
(a ) N
(a)
1) N
a 1
3
O a n ( a ) n ( a )
; F
a 1
(a)
O a n ( a ) N
a 1
1
1 (a ) (a)
N n
O
1 a
(8.87)
To see how the above relationships are proven, see the Section on Polar Decomposition in
Chapter 2.
Now, the second Piola-Kirchhoff stress tensor ( S ) in terms of the principal stretches
becomes:
S
w: (C )
wC
w: wO i
wO i wC
w: wO 1 w: wO 2 w: wO 3
2
wO 1 wC wO 2 wC wO 3 wC
(8.88)
Then by considering the spectral representation of C given in (8.82), the rate of change of
C (O a ) can be evaluated as follows:
C
2O
a 1
( a ) ( a )
( a )
(a ) N
( a ) O2 N
(a) N
N
N O2a N
a
aOa
(8.89)
2O
a 1
(a ) N
(a) N
(a) N
( a ) O2 N
(a) N
( a ) N
(a ) N
(a) N
(a ) N
(a) N
( a ) N
( a )
N
a
aOa
(8.90)
(a ) N
( a ) 1 , and the fact that the rate of change of a vector
Then, bearing in mind that N
with constant magnitude is always orthogonal to itself (Holzapfel (2000)), it follows that
(a) N
( a )
N
2O a O a
(8.91)
438
The reader should be aware here that the index a 1,2,3 is not a dummy index, i.e. we are
not dealing with indicial notation.
&
&
&
&
&
&
wC
O a we can obtain:
wO a
wC ( a )
(a)
:N N
2O a O a
wO a
1 (a)
( a ) 1 wC
:
N N
wO
2O a
2O a
1 (a)
( a )
:
N N
2O a
wC wO a
O a wC
w
(8.92)
1 (a) (a)
N N
2O a
(8.93)
Then, by using the second Piola-Kirchhoff stress tensor expression obtained in (8.88), i.e.
3
w: (O a ) wO k
w: wO a
, and by considering the equation obtained in (8.93), we
S 2
2
wO k
wC
a 1
wO a wC
1 w: ( a ) ( a )
N N
a wO a
a 1
S N
a
(a)
(a)
N
(8.94)
a 1
where S a are the second Piola-Kirchhoff stress tensor eigenvalues. Then, by comparing
the equation in (8.94) with the spectral representation of the tensor C , given in (8.82), we
can conclude that in isotropic materials, C and S are coaxial tensors ( C S S C ), i.e.
they have the same principal directions.
Then, as regards the Cauchy stress tensor, we have:
J 1 F S F T
1
a 1
J 1
a 1
1 w:
(a ) N
(a )
FN
O a wO a
1 w:
(a) N
(a ) F T
F N
O a wO a
J
3
1
a 1
1 w:
(a) F N
(a)
F N
O a wO a
(a)
Moreover, if we take into account the equation F N
Decomposition of F in Chapter 2), we can obtain:
J
a 1
1
Oa
w: ( a ) ( a )
n n
wO a
V n
a
(a )
(8.95)
n ( a )
(8.96)
a 1
w:
2
wI C
w:
w:
w:
II C
III C Oa2
III C Oa4
w
w
w
I
I
I
I
I
I
I
C
C
C
(8.97)
8 HYPERELASTICITY
439
F S
S N
a
(a )
(a)
N
a 1
Pa n
(a )
aF
N ( a ) N ( a )
a 1
(a )
(a) N
a O an
a 1
(8.98)
(a)
N
a 1
w:
2O a
wI C
w:
w:
w:
II C
III C Oa2
III C Oa4
w
w
w
I
I
I
I
I
I
I
C
C
C
(8.99)
We can also express these components in terms of Pa . To do so, let us consider that:
w: ( I C , II C , III C )
wO a
w: wI C
w: w II C
w: w III C
wI C wO a w II C wO a w III C wO a
(8.100)
wI C
wO 1
w
O21 O22 O23
wO 1
wI C
wO 2
w
O21 O22 O23
wO 2
wI C
wO 3
w
O21 O22 O23
wO 3
w II C
wO 1
2O >O O
2O >O O
2O 1
2O 2
2O 3
2 2
1 2
O21O23
2 2
1 2
O21O23
2
1
2O a
(8.101)
O @ 2O >O O O O O O O O O O @
O O O O O O O @ 2O II O III O
w
O21 O22 O22 O23 O21O23
wO 1
1
wI C
wO a
2O 1O22 2O 1O23
2 2
2 3
2
1
2 2
1 2
2O 1 O22 O23
2 2
1 3
2 2 2
2 3 1
2
1
2 2
2 3
4
1
2 2 2
2 3 1
2
1
(8.102)
4
1
w II C
2O a II C Oa2 III C Oa4
wO a
w III C w III C 2 2 2
O 1O 2 O 3 2O 1 O22 O23
wO 1
wO 1
(8.103)
>
2O 1 III C O12
(8.104)
2O a III C Oa2
(8.105)
w:
2O a
wI C
w:
w:
w:
II C
III C Oa2
III C Oa4
w III C
w II C
w II C
(8.106)
Additionally, if we compare (8.106) with (8.99) and with (8.97) we can draw the conclusion
that:
Pa
w: ( I C , II C , III C )
wO a
O aS a
(8.107)
440
(a)
(a) N
aS an
a 1
P n
a
(a)
(a)
N
(8.108)
a 1
Note that Pa are not the eigenvalues of P . The Cauchy stress tensor is related to the first
J 1 P F T , after which the eigenvalues of
J 1O a
w:
wO a
(8.109)
which is the same result as that obtained in (8.96). Note that index a does not indicate
summation.
Then, in isotropic materials, the Kirchhoff stress tensor ( ) and the left stretch tensor ( V )
have the same principal directions, and if we consider that J we can obtain:
2
(a)
a S an
n ( a )
a 1
W n
a
(a )
n ( a )
(8.110)
a 1
Now, if we look back at the equations in (8.94), (8.96) and (8.110), we can conclude that
the principal values of the tensors S , , , are interrelated by:
Sa
1 w:
O a wO a
J
Va
O2a
1
Wa
O2a
(8.111)
(8.112)
~
~
where F show an isochoric transformation ( F { F iso ), and F vol describes a dilatational
transformation, (see Figure 8.4). Now let us look back at Chapter 2 subsection 2.13, where
we obtained the following equations:
~
F
1
3 F
F vol
1
31
2
~
C J 3 C
2
~
3
b J b
;
;
C vol
J 31
vol
2
31
(8.113)
and
~
J
~
F
1
3 F
J vol
F vol
J 31
(8.114)
8 HYPERELASTICITY
w III C 3
wC
2
3
wJ
wC
obtain:
~
wC
wC
w III C
wC
1
J
3
2
3
III C C T
~
wC ij
2
J
J
2
3 I
2
3 PT
1
J
3
2
3 C
w( J
(8.115)
1
1
1
C C
3
2
3 C
ij
wC kl
wC kl
w (C )
w( J 3 )
C
wC
wC
2
3 I
4
1
III C 3 III C C T
3
2
III b . Likewise,
III C
C 1
w( J 3 C )
wC
2
3
J 1
C , where J 2
2
4 w III
1
C
III C 3
wC
3
1
1
III C 3 C 1
3
wJ
wC
441
2
3
2
3
2
3
w (C ij )
wC kl
E ik E jl
2
w( J 3 )
C ij
wC kl
1
J
3
2
3 C
(8.116)
1
ij C kl
1
I ijkl C ij C kl
3
with which, we introduce the fourth-order tensor P known as the projection tensor with
respect to the reference configuration, (see Holzapfel (2000)):
PT
b vol
1
I C C 1
3
F vol F vol
&
X
pure dilatation
F vol
C
B0
C
2
31
&
X
FT F
(8.117)
vol
1
I C 1 C
3
~
C
~
~
FT F
~
F
current
configuration
reference
configuration
F
~
F F vol
B
b
~
F
&
x
F FT
442
8.4.1
Next, we will define the stress tensors in different configurations. To start off we will use
2
w: (C )
F
FT
the definitions of the Cauchy stress tensor (current configuration)
and the second Piola-Kirchhoff stress tensor S
w: (C )
2
wC
wC
JF
1
T
(reference
configuration). Note that we can define a stress tensor, analogous to the Cauchy stress
tensor, in the intermediate configuration ( B ) by means of the transformation F vol , i.e.:
2
vol
vol
F vol
w: (Cvol ) F vol
vol
wJ vol
wC vol
J vol vol 1
C
2
vol
2
J vol
2
J
vol
2
J
vol
F vol
J vol
J
2
2
3 1
J 31 ,
w: (C vol )
wC vol
w: ( J vol ) wJ vol
, and
wJ vol wC vol
: (C vol )
(8.118)
wC
J , F vol
wC vol
F vol
2
J vol
1
w: ( J vol ) wJ vol
J 3 1
vol
wC vol
wJ
13
J 1
1
w: ( J vol ) J vol vol 1 13
J 1
C
J 3 1
vol
2
wJ
(8.119)
1
w: ( J vol ) J vol 2 1
J 3 1 J 3 1
J 31
wJ vol
2
w: ( J vol )
1
wJ vol
Thus,
vol
w: ( J )
1
wJ
(8.120)
We can also define a stress tensor in the intermediate configuration caused by the
~
transformation F , (see Figure 8.5), as:
~ ~
~
w: (C )
(8.121)
S 2
~
wC
We will now observe additive decomposition of the strain energy function in two parts,
namely: isochoric and volumetric, i.e.:
~ ~
~ ~
(8.122)
: ( F ) : ( F ) : vol ( F vol ) ; : (C ) : (C ) : vol (C vol )
Then, if we take the chain rule of derivative of the strain energy function (8.122) we obtain:
~ ~
~ ~
~
~ ~
w: (C ) wC d: vol ( J ) wJ w: (C ) ~ d: vol ( J )
(8.123)
: (C ) : (C ) : vol ( J )
J
~ :
~ :C
wC
wt
dJ
wt
wC
dJ
wJ J 1
J 1
~
wJ
C , we can obtain J
C : C and the term C
: C and
wC 2
2
wC
~
2
~ wC
can be expressed as C
: C J 3 P T : C . So, the equation in (8.123) can also be
wC
expressed as follows:
8 HYPERELASTICITY
443
~ ~
w: (C ) T J d: vol ( J ) 1
C :C
~ :P :C
dJ
2
wC
~ ~
2
w: (C ) J d: vol ( J ) 1
J 3 P:
C :C
~ :C
dJ
2
wC
: (C ) J
2
3
w: ( J )
1
wJ
vol
~
S
~
F
~
F
F vol
~
F F vol
~ ~
: (C ) : (C ) : vol (C vol )
w: (C )
wC
JF 1 F T
~
S S vol with S vol
vol
current
configuration
F
&
X
w: vol ( J ) 1 ~
C , S
wJ
2
3 P
~
J 1
~ ~ ~
: iso ( F ) { : ( F )
pure dilatation
reference
configuration
B0
~ ~
w: (C )
~
wC
&
X
: vol ( F vol )
(8.124)
w: ( J )
1
wJ
&
x
2
w: (C )
FT
F
J
wC
~
:S
Dint
1
S : C : (C ) 0
2
(8.125)
Then, if we combine the equation in (8.124) with the one above we obtain:
~ ~
2
w: (C ) d: vol ( J ) J 1
1
3
Dint
S :C J
C
~ :C
dJ
2
wC
~ ~
2
vol
S 2 J 3 P : w: (C ) J d: ( J ) C 1 : C
~
2
dJ
wC
P:
:C
(8.126)
0
444
Notice that the above must be satisfied for any admissible thermodynamic process. Let us
now consider that C z 0 , so, the only way for (8.126) to be satisfied is if:
~ ~
2
d: vol ( J ) 1
w: (C )
(8.127)
S 2J 3 P :
C
~ J
wC
dJ
Next, if we take the definition of the tensor S given in (8.11), and use the definition of
energy in (8.122), we can obtain:
~ ~
w: (C )
w: (C )
w: vol (C vol ) ~
w ~ ~
(8.128)
2
: (C ) : vol (C vol ) 2
2
S 2
S S vol
wC
wC
>
wC
wC
w: vol (C vol )
wC
w: vol ( J ) wJ
wJ
wC
~ ~
w: (C )
wC
~ ~
~
w: (C ) wC
~ :
wC
wC
w: vol ( J ) 1
JC 1
2
wJ
w: vol ( J ) 1
C
wJ
(8.129)
and
~
S
~ ~
w: (C )
~
wC
2
3 P :2
2
3 P
~
:S
(8.130)
~
where we have used the definition in (8.121). Additionally, we can verify that the tensor S
is in the intermediate configuration, i.e. it is only characterized by a change of shape, (see
Figure 8.5). Then, in summary we have:
~
S S S vol
(8.131)
where:
~
S
S vol
2
3 P
:2
~ ~
w: (C )
~
wC
d: vol ( J ) 1
C
dJ
2
3 P
~
:S
~
S
with
~
w: iso (C )
~
wC
(8.132)
JpC 1
(8.133)
In addition, with the constitutive equation for hydrostatic pressure, Holzapfel (2000):
D: vol ( J )
(8.134)
DJ
1
3
>x ( X& , t )@
dev
(x)
1
>(x) : C @C 1
3
(8.135)
Thus,
~
S
2
~
3 P:S
2
3
1 1
~
I 3 C C : S
>S~ @
dev
:C
2
3
~ 1 ~
1
S 3 S : C C
2
3
>S~ @
dev
(8.136)
(8.137)
8 HYPERELASTICITY
8.4.2
445
In compressible isotropic materials, the energy function decomposition can be given by:
~ ~
(8.138)
: (C ) : (b) : (b ) : vol ( J )
Then, as J
the following:
wJ
wb
J 1
b
2
wb~
wb
~
wbij
wb
kl
J
J
wJ
wC
J 1
C and (8.116) are valid, so are
2
2
3
1
I b b
3
2
3
1
E ik E jl bij bkl
3
(8.139)
Then, if we refer to the constitutive equation for isotropic hyperelastic materials obtained
w: (b)
in (8.70), J 1 2
b J 1 2b w: (b) , and incorporate the definition of energy
wb
wb
w: (b)
b
wb
~
vol
J 1 2
J 1 2
>
w ~ ~
: (b ) : vol ( J ) b
wb
J 1 2
~ ~
w: vol ( J )
w: (b )
b J 1 2
b (8.140)
wb
wb
J 1 2
w: vol ( J )
b
wb
J 1 2
w: vol ( J ) wJ
b
wJ
wb
J 1 2
w: vol ( J ) J 1
b b
2
wJ
w: vol ( J )
1 (8.141)
wJ
~ ~
w: (b )
1
I
:
b
b
~
3
wb
wb
~ ~
2
~
2
1
w: (b ) ~
b J 3 I b 1 J 3 b b 1 : 2 J 1
~ b
3
wb
~ ~
2
~ ~
w: (b ) ~
1
1
~
b J 3 I b 1 b b 1 : 2 J 1
~ b I 1 1 :
3
3
b
w
J 2b
J 2b J
(8.142)
>~ @dev
where we have used the relationship b J 3 b , and it can be proven that if x is a second
1
3
order tensor, then it holds that: I 1 1 : x x dev , where x dev represents the deviatoric
part of the tensor x , (see Problem 1.26).
Next, we will make the algebraic operations carried out in (8.142) using indicial notation:
446
~
V
ij
~ ~ wb~
w:
(b ) pq
J 1 2bik ~
wb
wbkj
pq
~ ~
2
w: (b )
J 1 2bik J 3 E pk E qj b pq bkj1 ~
3
wb pq
~ ~
2
1
w: (b )
J 1 2bik J 3 E pk E qj b pq bkj1 ~ E tp
3
wbtq
~ ~
2
1
w: (b ) ~ ~
J 1 2bik J 3 E pk E qj b pq bkj1 ~ bts bsp1
3
wbtq
J 1 2bik
~ ~
w: (b )
wbkj
(8.143)
E tp
~ ~
2
2
w:
~
(b ) ~
1~
J 1 2 J 3 bik E pk E qj bsp1 bsp1b pq J 3 bik bkj1 ~ bts
wb
3
tq
~ ~
~ ~
1~ ~
w: (b ) ~
J 1 2 bip bsp1E qj bsp1b pq bik bkj1 ~ bts
3
wbtq
~
~
1
1 w: (b ) ~
E is E qj 3 E ij E sq J 2 ~ bts
wbtq
~
vol
(8.144)
where
vol
w: vol ( J )
1
wJ
p1 ,
~
I 1 1 :
3
>~ @dev
(8.145)
8.4.2.1
D: vol ( J )
(8.146)
DJ
In isotropic materials, the strain energy function can be expressed in terms of the principal
invariants as:
~
~
: : ( I C~ , II C~ ) : vol ( J ) : ( I b~ , II b~ ) : vol ( J )
(8.147)
where : vol is a function of the third invariant of the right Cauchy-Green deformation
tensor ( III C J 2 ). This function, for an undeformed state, has to fulfill the following:
C
1
o III C
1
o: vol
w: vol
w III C
(8.148)
8 HYPERELASTICITY
I C~
2
3 I
IC
I b~
III C
II C~
4
3
II C
447
II C
3
II b~
III C2
III C~
III b~
(8.149)
Note that the invariants I C~ and II C~ are independent of the volumetric deformation.
We can now express the constitutive equation in the material description by means of S :
~
(8.150)
S S S vol
~
is the volumetric part, and S is the isochoric part, both of which are given
where S vol
respectively by:
~
d: vol ( J ) 1
C
JpC 1
; S
dJ
~ ~
w: (C )
2
~ can be demonstrated by:
wC
S vol
~
where S
~
S
w: ( I C~ , II C~ )
~
wC
2
3 P:2
~ ~
w: (C )
~
wC
2
3 P
~
:S
~
~
~
w:
w:
( I C~ , II C~ ) w: ( I C~ , II C~ )
( I C~ , II C~ ) ~
C
2
I C~ 1
wI C~
w II C~
w II C~
(8.151)
(8.152)
N 2
J 1
2
N
III C 1
2
(8.153)
and where N is the bulk modulus. This model at the limit J o 0 has no physical meaning
since : vol ( J o 0)
N
. Therefore, we can add the term : vol ( J )
2
N
log J 2 into the
2
equation, i.e.:
: vol ( J )
N 2
J 1 2 log J
4
(8.154)
which validated the following: when J o 0 , the energy function tends towards infinity, i.e.
: vol (J o 0) f . Additionally, in a small deformation regime J | 1 , the term 2 log J o 0 .
O 1O 2 O 3
J2
III C
III b
(8.155)
448
the volumetric deformation in incompressible materials is equal to zero for any hydrostatic
state. Note that here, even energy is not affected by the volumetric part, since:
: ( F vol ) : (1) 0 .
~
F,
pE ij
V ijhyd
p1
(8.156)
where p denotes pressure. Then, if we refer to the relationship between the Cauchy stress
and the second Piola-Kirchhoff stress tensors:
volumetric part of S :
S hyd
J F 1 hyd F T
J p F 1 1 F T
J p F 1 F T
J p B
J p C 1 (8.157)
C 1 , (see Chapter 2,
w: (C )
J p C 1
wC
(8.158)
: : (C ) H J 1
(8.159)
Now, the second Piola-Kirchhoff stress tensor can be obtained by taking the derivative of
the above equation with respect to C , i.e.:
S
w:
w
>: (C ) H J 1@ 2 w: (C ) 2H wJ
2
wC
wC
wC
wC
w: (C )
1
2
J HC
wC
w: (C )
J
2H C 1
wC
2
(8.160)
V kk
3
(8.161)
JF 1 F T
JF 1 ( dev V m 1) F T
S dev JV m C 1
S dev S hyd
JF 1 dev F T JF 1 V m 1 F T
(8.162)
Then, by comparing the result above with (8.160) we can clearly identify the deviatoric part
of the second Piola-Kirchhoff stress tensor.
After that, if we refer to the equation in (8.16) we can still write that in (8.160) as:
8 HYPERELASTICITY
F S
F 2
w: (C )
p F F 1 F T
wC
449
w: ( F )
p F T
wF
(8.163)
(8.164)
and:
w: ( F )
F T p F T F T
wF
PFT
w: ( F )
w: ( F )
F T p 1 F
p1
wF
wF
w: (C )
p C 1
wC
w: ( F )
p F T
wF
8.5.1
w: ( F )
w: ( F )
F T p 1 F
p1
wF
wF
The constitutive
equation for
hyperelastic
incompressible
materials
(8.165)
Geometrical Interpretation
In this section we will attempt to make a graphic interpretation of the results obtained
previously, (see Bonet&Wood(1997)). To start with, let us consider the internal energy
dissipation equation obtained in (8.9):
w: (C )
1
Dint S
:C 0
wC
(8.166)
As we saw before, in incompressible materials, the stress state is not completely defined by
the strain state. Then, the term written in parentheses in the equation in (8.166) is not equal
to zero, which indicates that C is not arbitrary, i.e. it has restrictions. Remember that the
J 1
C : C , (see Problem 2.12 in
2
Chapter 2). Additionally, if we consider the incompressibility condition J 1 , during
motion J 0 , we obtain:
J
J 1
C :C
2
(8.167)
The above equation gives us the restrictions on C which the equation in (8.166) has to
satisfy, thereby implying that:
w: (C )
1
S
wC
2
J 1
C (unit of stress)
2
(8.168)
& ( n )
w:
1
S
n t
w
2
C
J 1
C n
2
J & ( n ) 1
d
2
(8.169)
450
with which, the equations in (8.166) and (8.167) can be rewritten as:
&
&
t ( n ) d ( n )
&
1
&
J & ( n ) 1 & ( n )
d
d
0
2
&
are orthogonal to d (n ) , (see Figure 8.6).
(8.170)
&
t ( n )
J & ( n ) 1
d
2
J & ( n ) 1
d
2
n
&
&
d (n )
plane normal to n
8.5.2
: ( I C , II C ) : ( I b , II b )
(8.171)
: ( I E , II E )
(8.172)
w: ( I C , IIC )
w III C
p
wE
wE
w III C
w: wI C
w: w IIC
p
wI C wE w IIC wE
wE
(8.173)
Next, the relationships between the principal invariants of the tensors E and C (obtained
in Problem 2.10 in Chapter 2) are interrelated by:
IE
II E
III E
1
I C 3
I C 2I E 3
2
1
Reciprocal
2 I C II C 3
o II C 4 II E 4 I E 3
4
1
III C II C I C 1
III C 8 III E 4 II E 2 I E 1
8
(8.174)
8 HYPERELASTICITY
451
wI C
w
2 I E 3 2 1 , w II C w 4 II E 4 I E 3 4I E 1 E 4 1
wE wE
wE
wE
w III C
w
8 III E 4 II E 2 I E 1 8 III E E 1 4I E 1 E 2 1
wE
wE
(8.175)
>
w:
w:
>4I E 1 E 4 1@ p 8 III E E 1 4I E 1 E 2 1
21
w IIC
wI C
>
w:
w:
21
4>I E 1 1 E @ 2 p 2 I E 1 1 2 E 4 III E E 1
w IIC
wI C
(8.176)
1
2 II E I E
4
(8.177)
Then, by using the equation in (8.72), the constitutive equation for isotropic incompressible
materials becomes:
w:
w: 2
w:
I b b 2
b p1
2
w II b
wI b w II b
(8.178)
Note that we can still express the constitutive equation for incompressible hyperelastic
materials in terms of principal stretches, by means of the equation in (8.96):
Va
8.5.2.1
p Oa
w:
wO a
a 1,2,3
(8.179)
c I
pq
3
II C
3
(8.180)
p ,q 0
452
8.6.1
In the Neo-Hookean material model, the strain energy function is given in terms of the
isochoric part of C as follows:
N
where c1
N
2
I C
3 c1 I C 3
(8.181)
N
NT , where N is the number of polymer chains per unit of the
2
reference volume, N is the Boltzmann constant and T is the temperature. Lastly, the
parameter, N G can be determined experimentally and is known as the shear modulus.
Treloar (1944)), by N
Then, the stress constitutive equation for the Neo-Hookean material model becomes:
p1 2c1 b
8.6.2
(8.182)
This model expresses the strain energy function in terms of the principal stretches and is
given by:
: (O 1 , O 2 , O 3 )
Bp
p 1
O2 p O3 p 3
b >O O C O O C
N
q 1
O 2 O 3
Cq
(8.183)
3 hO 1O 2 O 3
This model expresses the strain energy function in terms of the principal stretches and is
given by:
N
Np
: (O 1 , O 2 , O 3 )
p 1
Bp
1
O2 p O3 p 3
(8.184)
where N , N p , B p are the material constants. In general, the shear modulus N , in the
reference configuration, becomes:
2N
Bp
N pB p ! 0
with
(8.185)
p 1
In the literature, e.g. Holzapfel (2000), we can find the following values for the constants
when p 3 :
; B 2 5.0
; B 3 2.0
B 1 1 .3
N1 6.3 u 10 5 N / m 2 ; N 2 0.012 u 10 5 N / m 2 ; N 3 0.1 u 10 5 N / m 2
(8.186)
N
2
2
1
O22 O23 3
N
2
I C
3
(8.187)
8 HYPERELASTICITY
453
N 1 and
(8.188)
Then, taking into account the equations in (8.11) and (8.67), the constitutive equation
becomes:
S
w: (C )
wC
wJ
2 a1 1 b1 II C C 1 III C C 2 h c( J )
wC
(8.189)
Afterwards, the derivative of the Jacobian determinant with respect to the tensor C can be
evaluated as follows:
wJ
wC
1
w
III C 2
wC
1 w III
1
C
III C 2
2
wC
1
1
III C 2 III C C 1
2
1
JC 1
2
(8.190)
8.6.3
2a1 1 b1 II C h c( J ) J C 1 b1 J 2 C 2
2
(8.191)
This model has the same energy expression as that provided by (8.184). Then, with the
parameter values N 2 , B 1 2 , B 2 2 , and with the constraint O21O22 O23 1
(incompressibility), the strain energy density given in (8.184) becomes:
: (O 1 , O 2 , O 3 )
N1
2
2
1
O22 O23 3
N2
2
2
1
O22 O32 3
(8.192)
I C~ and:
O22 O23 O21O23 O21O22
1
1
1
O21 O22 O23
O21O22 O23
IIC
III C
IIC
(8.193)
since we have the constraint III C 1 we can summarize the strain energy density as follows:
: (C )
where c1
N1
2
N1
2
and c 2
I C
3
N2
2
N2
2
IIC
3 c1 I C 3 c 2 IIC 3
IIC
(8.194)
3 ensure that the
strain energy is zero when there is no deformation ( E 0 ), since in this scenario and
according to the equation in (8.172), we will obtain I C 3 and IIC 3 . Note that in the
454
The second Piola-Kirchhoff stress tensor for the Mooney-Rivlin material model becomes:
S
w:
w:
w:
w:
C
2
I C 1
III C C 1
wI C w II C
w II C
w III C
w:
w:
w:
C 2>c1 c 2 I C 1 c 2 C @
2
I C 1
w
w
I
I
I
C
C
w II C
w: (C )
wC
(8.195)
8.6.4
p1 2c1b 2c 2 b 1
(8.196)
8.6.4.1
The Yeoh material model is used to simulate isotropic incompressible materials. Our
starting point here is the series expansion of strain energy density:
N
: : ( I C , II C , III C )
c I
pqr
3
II C
3
III C
1
(8.197)
p ,q ,r 1
N 3
c I
p
3
(8.198)
p 1
with which we can obtain the strain energy density for the Yeoh material model as follows:
: c1 I C 3 c 2 I C 32 c3 I C 33
(8.199)
8.6.5
w:
w:
w:
I C 1
2
I
I
I
w
w
C
w II C
w:
C
III C C 1
w
I
I
I
C
w: (C )
wC
w:
2
1 2 c1 2c 2 I C 3 3c3 I C 3 1
wI C
>
(8.200)
The Arruda-Boyce Material Model, also called the 8-chain model, takes into consideration
that the shear modulus, N , depends on the strain. This phenomenon is detected in some
polymers.
8 HYPERELASTICITY
455
: : (C ) N 0
p
cp
2 p2
1 M lock
p
C
3p
(8.201)
where N 0 is the initial shear modulus, ci are constants obtained by statistical theory and
M lock and N are material constants. Then, if we consider that N 3 we obtain:
: (C ) N 0 c1 I C 3
c2
2
C
c3
9
3
C
27
M
M
1
1
11
I C2 9
I C3 27
N0 IC 3
2
4
1050M lock
20M lock
2
8.6.6
2
lock
4
lock
(8.202)
1
N
N II
N
f I C 3
III CC 1 (1 f ) C 3 ( III CC 1)
2C
2 III C
2
C
: ( I C , II C , III C )
(8.203)
where C is given in terms of the N (shear modulus) and Q (Poissons ratio) by C
and f >0,1@ is an interpolation parameter. In the particular case in which f
: ( I C , II C , III C )
N
2
I C
3
N
2C
III
C
C
Q
,
1 2Q
1 , we obtain:
1
(8.204)
In the incompressibility case ( III C 1 ), the equation in (8.203) becomes the Mooney-Rivlin
model, (see equation (8.194)). Then, in the restrictive case where f 1 , and with the
incompressibility condition ( III C 1 ), we revert to the Neo-Hookean incompressible model
given in (8.181).
8.6.7
8.6.7.1
In the Saint Venant-Kirchhoff model, the strain energy density ( : ( I E , II E ) ) is given by:
: ( I E , II E )
1
M 2N I E2 2N II E
2
(8.205)
1
I C 3
2
II E
1
2 I C II C 3
4
III E
1
III C II C I C 1
8
(8.206)
456
: ( I C , II C )
8.6.7.2
1
1
M 2N I C 3 2N 2 I C II C 3
2
2
N
1
M 2N I C 32 2 I C II C 3
8
2
(8.207)
The derivatives of the function (8.205) with respect to the invariants are:
w:
wI E
M 2NI
w:
w II E
2N
(8.208)
w:
w:
w:
IE
II E
wI E w II E
w III E
c1
w:
w:
IE
w II E w III E
M 2NI
2NI E
OI E
(8.209)
2N
and by substituting the above parameters into the equation in (8.79), we obtain:
S MI E 1 2NE
(8.210)
1
2
C 1 and
IE
1
2
I C
M
M 12 I C 31 2N 12 C 1
I C
3 N 1 NC
(8.211)
Note that the above equation could have been obtained by means of the constitutive
equation in terms of S given in (8.67), i.e.:
S
w: (C )
wC
w:
w:
w:
2
I C 1
wI C w II C
w II C
w:
w:
w:
2
I C 1
w II C
wI C w II C
w:
C
III C C 1
w III C
N N
N
2
2 M 2N I C 3 I C 31 C (8.212)
2 2
2
8
I C 3 N 1 NC
2
8.6.7.3
wS
wE
w
w
M I E 1 2N E {
M I E 1 2N E
wE
wE
wI E
wE
1 2N
wE
wE
(8.213)
8 HYPERELASTICITY
457
Note that E is a symmetric tensor, so the result of the operation E , E is also a symmetric
tensor. Then, the equation in (8.213) in indicial notation becomes:
C tan
ijkl
wS ij
wE kl
wE ij
wI E
E ij 2N
wE kl
wE kl
w
wI E
E ij 2N
wE kl
> E
1
2
ij
(8.214)
M 1 1 2N I
(8.215)
E ji
wE kl
Thus,
ME kl E ij N>E ik E jl E il E jk @
C tan
ijkl
notation
tensorial
o
C tan
NOTE: Note that in a small deformation regime, the condition that all stress tensors are
equal is satisfied, i.e. S | , and the same is true for the strain tensors, E | , after which
the constitutive equation for stress in (8.210) becomes:
M I 1 2N
(8.216)
which is the same constitutive equation for isotropic linear elastic materials as that obtained
in Chapter 7, (see also Problem 6.1). In addition, the elastic tangent stiffness tensor C tan
coincides with the elasticity tensor C e , (see Chapter 7).
8.6.8
8.6.8.1
In the compressible Neo-Hookean material model, the Helmholtz free energy per unit
reference volume (strain energy density), (see Bonet&Wood (1997)), is defined by:
: (C )
( lnJ ) 2 N lnJ ( I C 3) : ( J ) : ( I C )
2
2
: (J )
(8.217)
: ( IC )
The second Piola-Kirchhoff stress tensor, (see Eq. (8.67)), is given by:
S
w: (C )
wC
w:
2
wI C
w: ( I C , II C , III C ) w: ( J )
2
wC
wC
w:
w:
1
II C
III C
w III C
w II C
1 w:
w: ( J )
C
III C C 2 2
wC
w II C
(8.218)
1 wJ
11
w:
w:
w: ( J ) M
0,
0 and
N
2( lnJ )
JC 1 . Moreover,
J wC
J 2
2
wC
w II C
w III C
wJ J 1
if we consider that
C , we can conclude that:
wC 2
where
w:
wI C
N
N M
2 1 2 ( lnJ )C 1 C 1 N 1 C 1 M ( lnJ )C 1
2
2 2
(8.219)
Then, by considering the relationship between the Cauchy stress tensor and the second
Piola-Kirchhoff stress tensor, J F S F T , as well as C 1 F 1 F T , we can obtain:
458
>
@
@ F
F S F T
F N 1 C 1 M ( lnJ )C 1 F T
F N1F
1
>
NF F
T
F M ( lnJ ) F F
NF F 1 F T F T M ( lnJ ) F F 1 F T F T
T
1
T
(8.220)
Thus,
>
1
Nb 1 M ( lnJ )1
J
8.6.8.3
(8.221)
The elastic tangent stiffness tensor, (see Figure 8.3), can be defined as:
C tan
w 2:
wC wC
wS
wC
(8.222)
Then, given the second Piola-Kirchhoff stress tensor equation in (8.219), we obtain:
>
w
N 1 C 1 M ( lnJ )C 1
wC
w 1
w ( lnJ )
wC 1
wC 1
2 N
M (ln J )
N
M C 1
wC
wC
wC
wC
C tan
In Chapter 1 we obtained
wC ir1
wC kl
>
(8.223)
1 1 1
C ik C lr C il1C kr1 , after which the above equation,
2
2N
wC ir1
wC 1
w ( lnJ )
2M C ir1
2M ( lnJ ) ir
wC kl
wC kl
wC kl
C ik1C lr1
C il1C kr1
2M
C ir1
1 J 1
C kl M ( lnJ ) C ik1C lr1 C il1C kr1
J 2
(8.224)
>N M ( lnJ )@ C C
>N M ( lnJ )@ C C
1
ik
C tan
irkl
1
C tan
1
lr
1
C 1 C 1 M C 1 C 1
(8.225)
We can also obtain the tensor L , (see Figure 8.3), which is related to the tensor C by the
equation in (8.32), i.e.:
L abcd
^>
1 1
1 1
Fbq Fap N M ( lnJ ) C ps
C tq C pt1C sq1 MC pq
C st Fct Fds
1
Fbq Fap F px1 Fsx1 Ftw1 Fqw
Fct Fds
1 1
E ac E bd , Fbq Fap C pq
C st Fct Fds
E ax E bw E xd E wc
(8.226)
E ad E bc ,
E ab E cd .
Then:
L abcd
>N M ( lnJ )@ E
ad
E bc E ac E bd M E ab E cd
(8.227)
sym
2I abcd , where I { I sym is the symmetric fourth-order
Note that E ad E bc E ac E bd 2I abcd
unit tensor, with which we can obtain:
8 HYPERELASTICITY
L abcd
>
459
>
2 N M ( lnJ ) I M1 1 (8.228)
Tensorial notation
2 N M ( lnJ ) I sym
abcd OE ab E cd o L
Then, if N c
1
L
J
>N M ( lnJ )@
J
>N M ( lnJ )@
J
and M c
I
M
J
(8.229)
1 1
(8.230)
dx1
dx
2
dx 3
dV
O 0 0 dX 1
0 O 0 dX
2
0 0 O dX 3
Stretches O O 1
O2
O dX 3
dX 3
O3
dV 0
dX 1
OdX 1
dX 2
X1
X2
O dX 2
O J3
The Cauchy stress tensor can be obtained by means of the equation in (8.221), i.e.:
>
1
N b 1 M ( lnJ )1
J
O1 O1
O2 1
J 31
N 2
M
J 3 1 ( lnJ ) 1
460
8.6.9
Strain energy density in the Gent model, Gent(1996), is characterized by the following
logarithmic function:
: (I C )
I 3
N
I m ln1 C
2
I m
(8.231)
where N and I m are material constants, and I m is the constant that measures the limit
value of I C 3 .
~
Let : be the isochoric part and : vol the volumetric part, then, the energy function is
given by:
~
: : ( I C~ ) : vol ( III C )
(8.232)
where
~
1
1
I 2~ 9 11 2 I C3~ 27
: N I C~ 3
20 N C
1050 N
2
: vol
N
ln III C
2
(8.233)
519
19
I C4~ 81
I C5~ 243
4
3
673750 N
7000 N
(8.234)
a 0 1 a 2 C 1
(8.235)
1
11
19
519
1
1
I C~
I 2~
I 3~
I 4~
2N
4 C
2 C
3 C
2
10
N
350 N
1750 N
134750 N
3 III C
(8.236)
2 1
1 2
11
19
I~
I 3~
I 4~
N I C~
3 2
10 N C 350 N 2 C 1750 N 3 C
519
I 5~ N ln III C
134750 N 4 C
(8.237)
C tan
ijkl
a 2 1 1
C ik C lj C il1C kj1
2
(8.238)
11
57
1038
1
1
I~
I 2~
I 3~
2N
2
3 C
67375 N 4 C
10 N 175 N C 1750 N
3 III C
3 III
C
(8.239)
8 HYPERELASTICITY
461
2 1
1
33
38
519
1
N
I~
I 2~
I 3~
I 4~
3 2 5 N C 350 N 2 C 875 N 3 C 26950 N 4 C
3 III C
b2
b4
b5
a2
2 1
1 2
33
38
519
N I ~
I~
I 3~
I 4~
I 5~
9 2 C 5 N C 350 N 2 C 875 N 3 C 26950 N 4 C
N
2
2 1
1 2
11
19
I~
I 3~
I 4~
N I C~
3 2
10 N C 350 N 2 C 1750 N 3 C
519
I 5~ N ln III C
134750 N 4 C
(8.240)
(8.241)
(8.242)
(8.243)
The demonstration of the statistical model can be found in Chaves (2009), (see also
Sansour et al. (2003)).
~
: : ( I C~ , II C~ ) : vol ( III C )
(8.244)
where
~
: ( I C~ , II C~ ) B 1 I C~ B 2 II C~ B 3 I C2~ B 4 I C~ II C~ B 5 I C3~ B 6 II C2~ B 7 I C4~ B 8 I C~ II C2~
:
B 9 ln III C
vol
(8.245)
(8.246)
The parameters B 1 , B 2 , ..., B 8 are the eight material parameters whereas B 9 represents
the bulk modulus. The values these parameters were determined by Sansour (1998) as:
B 1 0.1796;
B2
4
B 5 0.3473 u 10 ; B 6
B3
0.0145;
3
0.8439 u 10 ; B 7
0.1684 u 10 2 ; B 4
0.432 u 10 7 ;
B8
0.3268 u 10 3
(8.247)
0.5513 u 10 5
(8.248)
where
a0
2
III C
B 2 I C~ 2B 3 I C~ B 4 II C~ B 4 I C2~ 3B 5 I C2~
2B 6 I C~ II C~ 4B
a1
2 B 2 B 4 I C~ 2B 6 II C~ 2B 8 I C~ II C~
a2
3
7 I C~
B
2
8 II C~
2B
2
8 I C~
II C~
1
3
(8.250)
III C2
2
B 1 I C~ 2B 3 I C2~ 3B 4 I C~ II C~ 3B 5 I C3~ 4B 7 I C4~ 5B 8 I C~ II C2~
3
2B 2 II C~ 4B 6 II C2~ 3B 9 ln( III C )
(8.249)
(8.251)
462
b0 E ij E kl b1 E ij C kl C ij E kl b2 E ij C kl1 C ij1E kl b4 C ij C kl
C tan
ijkl
a1
a
E ik E jl E jk E il b5 C ij1C kl C ij C kl1 b8 C ij1C kl1 2 C ik1C lj1 C il1C kj1
2
2
(8.252)
where
b0
b1
b2
2
3
III C2
2B 3 3B 4 I C~ 6B 5 I C~ 2B 6 II C~ 2B 6 I C~2
2
B 4 2B 6 I C~ 2B 8 II C~ 2B 8 I C2~
III C
2 1
B 1 2B 2 I C~ 4B 3 I C~ 3B 4 I C2~ 3B 4 II C~ 9B 5 I C2~ 8B 6 I C~ II C~
3 3 III C
b8
(8.254)
16B
b4
(8.253)
2
3
III C4
2B
2B 8 I C~ , b5
2
3
1
3
III C2
2B
3
7 I C~
5B
2
8 II C~
10B
2
8 I C~
II C~
3B 4 I C~ 8B 6 II C~ 10B 8 I C~ II C~
2
B 1 I C~ 4B 2 II C~ 4B 3 I C2~ 9B 4 I C~ II C~ 9B 5 I C3~ 16B 6 II C2~
9
16B 7 I C4~ 25B 8 I C~ II C2~ 9B 9
(8.255)
(8.256)
(8.257)
The demonstration of the eight-parameter model can be found in Chaves (2009), (see also
Sansour et al. (2003)).
a 0
a 0
b 0
8 HYPERELASTICITY
8.7.1
463
(8.258)
where I C( 4) and I C(5) are the pseudo-invariants of anisotropy. Moreover, if we consider that
the energy, if independent of the sense of the vector a 0 , fulfills that : (C , a 0 ) : (C ,a 0 ) ,
we can then represent the strain energy function by:
: : (C , a 0 a 0 )
(8.259)
(8.260)
: (C , a 0 a 0 ) : (Q C Q T , Q a 0 Q a 0 ) : (C * , a *0 a *0 )
(8.261)
I C( 4)
I C( 4)
where I C( 4)
a 0 C a 0
;
;
I C(5)
I C(5)
a 0 C 2 a 0
a 0 i C ip C pj a 0 j
(8.262)
stretching).
So, the energy function can be expressed as follows:
: : ( I C , II C , III C , I C( 4) , I C(5) )
(8.263)
Now, the derivatives of I C( 4) and I C(5) with respect to the tensor C are given by:
wC ij
wI C( 4 )
wC kl
w
a 0 i C ij a 0 j
wC kl
wI C(5)
wC kl
wC ip
wC pj
w
a 0 i C ip C pj a 0 j a 0 i a 0 j C pj
a 0 i C ip a 0 j
wC kl
wC kl
wC kl
a 0 i a 0 j C pj E ik E pl a 0 i C ip a 0 j E pk E jl a 0 k a 0 j C lj a 0 i C ik a 0 l
a 0 i a 0 j
wC kl
a 0 i a 0 j E ik E jl
a 0 k a 0 l
(8.264)
a 0 a 0
wI C(5)
wC
a 0 C a 0 a 0 C a 0
(8.265)
464
Then, the constitutive equation for stress (reference configuration) can be represented by:
S
w: ( I C , II C , III C , I C( 4 ) , I C(5) )
wC
( 4)
( 5)
w: wI C
w: w II C
w: w III C
w: wI
w: wI C
( 4) C
2
wI wC w II wC
w III C wC
wI C wC
wI C wC
C
C
2
(8.266)
Next, if we consider the derivatives in (8.264) and the equation in (8.67) we obtain:
S
w:
2
wI C
w:
w:
w:
1
II C
III C C 1
III C C 2
w
I
I
I
I
I
I
I
w
w
C
C
C
w:
w:
a 0 a 0 (5) a 0 (C a 0 ) (a 0 C ) a 0
( 4)
wI C
wI C
(8.267)
w 2 I C( 4)
wC ij wC kl
0 ijkl
(8.268)
Note that the second derivative of I C(5) becomes a fourth-order tensor that features both
major and minor symmetry:
w 2 I C(5)
wC wC
w 2 I C(5)
wC ij wC kl
a 0 a 0 1
a 0 i a 0 j E kl
(8.269)
If we now consider the directions of the fibers in the current configuration (deformed
configuration), a , and using the expression of the Kirchhoff stress tensor, (8.71), we
obtain:
F S F T
w:
2F
wI C
w:
w:
w:
1
II C
III C C 1
III C C 2
I
I
I
I
I
I
I
w
w
w
C
(8.270)
w:
w:
( 4 ) a 0 a 0 (5) a 0 (C a 0 ) (a 0 C ) a 0 F T
wI C
wI C
FT F , b
F F T and F a 0
then becomes:
w:
w:
w:
w:
b2
I b b
III b 1
J 2
w II b
w III b
wI b w II b
w:
w:
a (b a ) (a b) a
O a
a a O2a
0 wI ( 4 )
0 wI ( 5 )
b
b
(8.271)
w:
a 0 C a 0
w:
O2a , we obtain:
0
w:
b2
I b b
III b 1
J 2
w II b
w III b
wI b w II b
I C( 4 )
w:
w:
a a I C( 4 ) (5) a (b a ) (a b) a
(4)
wI b
wI b
(8.272)
9 Plasticity
Plasticity
9.1 Introduction
With elastic loading, atomic structures are not affected, which is typical of processes in
which there is no internal energy dissipation. Then, once the load has been removed, the
solid returns to its initial state. In certain types of materials, if we keep loading, a level will
be reached in which the atoms begins to restructure (dislocation at the atomic level), so, in
this way, we have internal energy dissipation (an irreversible process). Most of the
dissipated energy will be used to increase the temperature (heat release), and as a result
there will be an increase in system disorder, i.e. a rise in entropy. A rise in temperature also
involves the dilation phenomenon. At the macroscopic level, in ductile materials, this
atomic restructuring is characterized by permanent deformation (plastic strain). That is, if
the material which has been internally restructured is completely unloaded, it can be
observed that part of the total deformation is regained. This recoverable part is
characterized by the elastic strain, and the permanent deformation by the plastic strain, (see
Figure 9.1), and the constitutive models devised to represent this phenomenon are called
plasticity models or elastoplastic models.
It can be complex to formulate a constitutive model that considers all possible phenomena
that occur during the plasticity process. In general, a process which involves plastic strain
typically involves a large deformation, heat production, and the loss of material isotropy in
the plastic zone due to plastic fibers which are formed in this area. However, in certain
types of materials, the effect of temperature can be discarded (isothermal process), and they
can also be subjected to a deformation state in which the elastic strain is very small when
compared with the plastic strain (small deformation regime). These simplifications have given
rise to the classical Theory of Plasticity.
465
466
VY
I - elastic zone
II - plastification zone
III - complete unloading
VY
III
II
II
H p - plastic strain
VY
H e - elastic strain
H
H
9 PLASTICITY
467
Continuum Mechanics
Solids
Fluids
Multiphysics
Kinematics
Constitutive Law
Plasticity
Viscous models
Linear Elasticity
Hyperelasticity
Hyperplasticity
Damage models, ...
Structural Mechanics
Figure 9.2: Overview of solids mechanics.
468
Metaphorically speaking, we can state that the constitutive equation reflects the
personality of the material, i.e. each material (or kind of material) has its own yield
surface and hence the concept of yield criterion appears. That is, the yield criterion establishes
when certain materials start to plastify.
9.2.1
F (V ij )
(9.1)
which represent the hypersurface equation in the stress space (six dimensions), in which:
F (V ij ) 0
elastic domain
F (V ij )
plastic domain
(9.2)
Alternatively, the equation in (9.1) can be rewritten in terms of the eigenvalues ( V a ) and
&
eigenvectors ( n ( a ) ) of , i.e.:
&
&
&
(9.3)
As expected, determining the yield criterion for anisotropic material is rather difficult to
achieve. Moreover, because of anisotropy, we need to establish 21 independent constants
in the laboratory (mechanical properties) in order to fully describe elastic material behavior.
9.2.1.1
wF ( )
w
components
o
n ij
wF ( )
wV ij
(9.4)
where n is a symmetric second-order tensor or the plastic flow tensor, with six independent
components as well as being a coaxial tensor with .
Then, the Frobenius norm of the tensor n is defined as follows:
F
F
with
F : F
n:n
(9.5)
n:n
1.
In isotropic materials the yield surface is only dependent on the -eigenvalues, hence, it
can be shown in the space defined by the principal stresses (three dimensions), so the
plastic flow becomes a vector, and because of this some researchers refer to n as the flow
plastic vector.
9.2.2
9 PLASTICITY
469
isotropic materials the beginning of plastification does not depend on the principal
directions (eigenvectors) of the Cauchy stress tensor, thus:
F (V 1 , V 2 , V 3 )
F ( I , II ,
III ) 0
(9.6)
where I , II , III are the Cauchy stress tensor principal invariants, whose values are
Tr ( ) , II
given by I
1
>Tr( )@ 2 Tr( 2 ) , III
2
(9.6) represent the yield surface in the space defined by the Cauchy stress tensor
eigenvectors. As discussed in the chapter on stress, (see Problem 3.3), the principal
invariants are related to the invariants of the deviatoric part of as follows:
J2
where J1
I dev
0 , J2
II dev
1
3 II I 2
3
J3
1
2 I 3 9 I II 27 III
27
1 dev
: dev , J 3
2
III dev
(9.7)
F (I , J 2 , J3 )
(9.8)
F (J 2 , J3 )
(9.9)
In this case, the yield surface is represented by a prism in the principal stress space, (see
Figure 9.3).
NOTE: Before starting to study the yield criteria, a review of subsection A.4 Graphical
Representation of the Spherical and Deviatoric Parts in Appendix A is recommended.
n { F
V1
hydrostatic axis
V3
V2
470
In Appendix A we saw that a stress state in the principal stress space (Haigh-Westergaard
stress space) can be represented by means of three variables ( p , q , R ) called the HaighWestergaard coordinates, (see Figure 9.4). Thus, the yield surface for isotropic materials
can also be shown as:
F ( p, q, R )
(9.10)
F ( q, R )
(9.11)
hydrostatic axis
3c
V1
P ( V1 , V 2 , V 3 )
V1
AV m , V m , V m
1
ec
2
ec
V3
n i {
>
1
3
1
3
1
3
p
D
V2
n
D
Deviatoric plane
3
ec
(octahedral plane)
V3
V2
Deviatoric 3 -plane
(Nadais plane)
S
.
3
Vc1
V1
V2
V1 ! V 3 ! V 2
V1 ! V 2 ! V 3
V1
V3
3
V 2 ! V1 ! V 3
V 3 ! V1 ! V 2
Vc2
Vc3
V 3 ! V 2 ! V1
V 2 ! V 3 ! V1
V2
V3
9 PLASTICITY
9.2.3
471
In the previous section we saw that, generally speaking, for materials in which the yield
surface is independent of the hydrostatic pressure, the former has a prismatic shape in
which the prismatic and hydrostatic axes coincide. The yield surface cross section is then
defined depending on the model adopted. In this subsection, we will introduce some
models developed to represent material plastification where the hydrostatic pressure has no
influence on plastification, (see Chen&Han(1988)).
9.2.3.1
The yield criterion of von Mises (1913) assumes that plastification occurs when the second
invariant of the deviatoric tensor, J 2 , reaches a critical value k 2 . We can define this
plastification criterion as:
J2 k 2 0
elastic domain
J2 k 2
plastic domain
(9.12)
where k is a material property (yield stress in pure shear). Remember that in Chapter 1 (see
also Problem 3.3), the second invariant J 2 can be written in terms of the Cauchy stress
tensor components. Thus, the yield surface for von Mises criterion can be written as:
>
1
2
2
(V11 V 22 ) 2 (V 22 V 33 ) 2 (V11 V 33 ) 2 V12
V 223 V13
k2
6
(9.13)
J2
>
1
(V 1 V 2 ) 2 (V 2 V 3 ) 2 ( V 1 V 3 ) 2 k 2
6
1 2
V1 V 22 V 32 V1V 2 V 2 V 3 V1V 3 k 2
3
(9.14)
The von Mises surface is defined by a cylinder, (see Figure 9.6), which prismatic axis is
parallel to hydrostatic axis, and does not depend on I nor R . We can obtain the cylinder
radius by r q
2J 2
2k.
2k
V1
V1
von Mises
V3
V2
V3
V2
Figure 9.6: Yield surface for von Mises yield criterion (independent of pressure).
472
The parameter k is obtained by means of tensile testing, where it holds that V1 V Y and
V 2 V 3 0 . Under these conditions the equation in (9.14) becomes:
V Y2
3k 2
VY
(9.15)
The strain energy density (the energy per unit volume) is given by:
:
1
V ij H ij
2
1 dev 1
dev 1
V ij V kk E ij H ij H pp E ij
2
3
3
(9.16)
1 dev dev 1
: Tr ( ) Tr ( ) : dev : vol
2
6
1 dev dev
V ij H ij
2
1 dev dev
V ij V ij
4N
1
J2
2N
(9.17)
Thus, it is possible to interpret the von Mises yield criterion as: plastification begins when
the energy : dev reaches the critical value
J2
2N
k2
2N
V Y2
.
6N
Another interpretation of the von Mises yield criterion is related to the octahedral shear
stress:
2
W oct
J2
>
1
(V 1 V 2 ) 2 (V 2 V 3 ) 2 (V 1 V 3 ) 2
9
1
(V 1 V 2 ) 2 ( V 2 V 3 ) 2 (V 1 V 3 ) 2
6
>
W oct
2
J2
3
(9.18)
whose equations were obtained in Appendix A. Then, by applying the von Mises yield
criterion (9.12), J 2 k 2 0 , we obtain:
2
W oct
2
J2
3
2 2
k
3
W oct
2
k
3
(9.19)
This criterion can also be interpreted as that in which material starts to plastify when the
octahedral shear stress reaches the critical value
2
k
3
VY
3
2.
Now, by considering the equations in (9.14) and (9.15) we can still write the von Mises
yield criterion as:
(V1 V 3 ) 2 (V1 V 3 )(V 2 V 3 ) (V 2 V 3 ) 2
V Y2
(9.20)
9 PLASTICITY
473
which shows the ellipse equation in the coordinate system (V1 V 3 ) - (V 2 V 3 ) , whose
ellipse axis form a 45 angle to the axes (V1 V 3 ) and (V 2 V 3 ) , (see Figure 9.7).
(V 2 V 3 )
45
VY
3
(V1 V 3 )
VY
VY
3
V1 V 2 V 22
3k 2
V Y2
(9.21)
which represents an ellipse in the V1 V 2 -space, (see Figure 9.8), and in uniaxial cases the
yield surface is reduced to a point.
V2
VY
uniaxial
VY
V1
pure shear
Figure 9.8: The von Mises yield criterion the state of plane stress.
The von Mises yield criterion can also be expressed in terms of the Frobenius norm of the
Cauchy deviatoric stress tensor ( dev { s ), which is given by:
s
s:s
s ij s ij
2 2
V1 V 22 V 32 V1 V 2 V 2 V 3 V1 V 3
3
2J 2
(9.22)
Then, by considering the equations (9.14), (9.15) and (9.22), the von Mises yield criterion
can be expressed as follows:
3
2J 2 VY
2
3
s VY
2
(9.23)
We can then summarize the different ways of expressing the yield surface for the von
Mises yield criterion:
474
F (J 2 , k)
J2 k 2
W oct
2
k
3
3J 2 V Y
F (W oct , k )
F (J 2 , VY )
F (W oct , V Y )
F ( s , VY )
3
s VY
2
W oct V Y
(9.24)
Next, we will calculate the yield surface gradient for the von Mises criterion in the principal
stress space. Starting with the definition n { F {
3J 2 V Y
wF
w
0 , we obtain:
wF
wV ij
n ij
w ( 3J 2 V Y )
w ( 3J 2 )
wV ij
wV ij
1
1
3J 2 2 w(3J 2 )
wV ij
2
(9.25)
1 2
V1 V 22 V 32 V1 V 2 V 2 V 3 V1 V 3 , and because we are
3
working in the principal stress space ( V11 V1 , V 22 V 2 , V 33 V 3 , V12 V13 V 23 0 ) we
2V 1 V 2 V 3
w(J 2 )
wV ij
2V 3 V 1 V 2
2V 2 V1 V 3
3
0
s ij
(9.26)
1
wF
wV ij
w (3J 2 )
1
(3J 2 ) 2
wV ij
2
3 s ij
2 2J 2
3 1
s ij
2 3J 2
3
2
s ij
s pq s pq
(9.27)
s:s
where s
wF
w
wF
wV ij
s
s :s
3 s
2 s
3
s
2
(9.28)
1 holds. The same result as in (9.28) could have been obtained starting
n ij
3
2
3
s VY
w
2
wV ij
3 s ws
:
2 s w
3
s VY
2
3 w s
2 wV ij
0 , i.e.:
3 w s w s ij
2 ws ij wV kl
3 s w s ij
2 s wV kl
(9.29)
9 PLASTICITY
ws
w
475
3 s
and
2 s
the reader will be left to verify that the tensors and n { F are coaxial, i.e.
F F . In other words, they have the same principal directions (eigenvectors).
9.2.3.2
In the Tresca yield criterion (1864) (also called the criterion of maximum shear stress) plastification
of the material begins when the maximum shear stress reaches the critical value k T .
Mathematically, this criterion is represented by:
W max k T
elastic domain
W max
plastic domain
kT
(9.30)
2
2
2
kT
(9.31)
Then, by considering the principal stresses such as V I ! V II ! V III , the Tresca yield
criterion is given by:
1
V I V III k T
2
1
V I V III k T
2
F (V I , V III , k T )
(9.32)
where we have considered that V I V III ! 0 . Note that the principal stress V II has no
influence on the Tresca criterion. Then, the material constant k T is evaluated by tensile
testing, where it holds that V I V Y , V III 0 , thus:
V1
VY
VY
2
kT
(9.33)
Figure 9.9 represents the stress state, at a material point, by means of a Mohrs circle in
stress. In this figure we can appreciate the elastic stress state (before plastification), and we
can verify the evolution of the stress sates from elastic to the initiation of plastification, (see
Figure 9.9).
W
Plastification zone
kT
Evolution
W max
W max
V II
VI
V III
V II
VI
VN
476
V1 V 2
VY
V1 V 2
V Y
V 2 V3
V 2 V3
VY
V 3 V1
V Y
VY
V 3 V1
(9.34)
V Y
where we have considered the equation in (9.33). Note that each one represents a plane
equation which is parallel to the hydrostatic axis. The surface generated by putting these
planes together is a prism whose cross-section is defined by a regular hexagon, (see Figure
9.10). As we can verify, the cross section of the prism is the same in shape and size for any
point on the hydrostatic axis.
hydrostatic axis
von Mises
Vc1
V1
Vc1
Tresca
V3
Vc2
3
V2
V1 ! V 2 ! V3
Vc3
0
0
V 3
0
V2
0
V m
0
0
Vm
0
cos R
0
0
2
J2 0
cos(R
0
3
0
0
V m
cos(R 23S )
0
2S
)
3
(9.35)
Then, by substituting the stresses V1 and V 3 from then above equation into the yield
criterion in (9.32) we obtain:
1
2
2
2
V m
J 2 cos R V m
J 2 cos(R
3
3
1
>cos R cos(R 23S )@ k T
J2
3
J 2 sin(R S3 ) k T
1
2
2S
)
3
kT
(9.36)
(9.37)
q sin(R S3 ) k T
(9.38)
F ( q, R )
2q sin(R S3 ) 2 k T
(0 d R d S3 )
(9.39)
9 PLASTICITY
477
As expected, we can see that the yield surface is not a function of the hydrostatic pressure
p.
S
, q
3
S
, q
6
Vc1
2 kT
R 0, q 2
2
kT
3
2
kT
3
Vc3
Vc2
Figure 9.11: Tresca yield surface Projection of the surface onto the deviatoric plane.
As for the state of plane stress, (V 3
V1 V 2
VY
V1 V 2
V Y
V2
;
;
V2
VY
V Y
V1
;
;
VY
V1
V Y
(9.40)
which is represented by a hexagon in the principal stress space V1 V 2 , (see Figure 9.12).
V I V II
r2k T
V I
V II
r2k T
r2k T
(9.41)
In Figure 9.13 we can see the von Mises and Tresca criteria for the state of plane stress
( V 3 0 ), which show that the Tresca hexagon is inscribed into the ellipse of von Mises.
The yield surface gradient for the Tresca yield criterion can be obtained by means of the
definition n { F {
wF
. Then, we can use the definition of the Tresca yield surface
w
1
V I V III k T F (V I , V III , k T ) 1 V I V III k T 0
2
2
1
12 0 0
w V I V III k T
1 w>V I V III @
wF
2
n ij
0 0 0
2
wV ij
wV ij
wV ij
0 0 21
9.2.4
(9.42)
(9.43)
Porous materials, e.g. soil, rock, concrete, and some porous metals, are affected by
hydrostatic pressure, i.e. these materials are dependent on the first invariant I , (see
Chen&Han(1988)). Here, can mention some criteria that take into account this
phenomenon, namely: the Mohr-Coulomb criterion, the Drucker-Prager criterion and the
Rankine criterion, among others.
478
V2
V1 V 2
VY
V Y
V2
VY
VY
V1
2k T
VY
VY
VY
V2
V Y
V1
V1 V 2
VY
VY
V2
VY
VY
VY
V1
VY
Figure 9.13: The yield curves for von Mises and de Tresca Plane stress.
9.2.4.1
This criterion was formulated by Coulomb in 1773, and was enhanced by Mohr in 1882,
Oller (2001). Mathematically, the Mohr-Coulomb criterion is given by:
W W(V i , k1 , k 2 , )
(9.44)
where (k1 , k 2 , ) are material constants and the function W(V i ) is obtained by means of
laboratory experiments; (see the Triaxial Compression Test in subsection 6.5.12 in Chapter 6).
The function W(V i ) corresponds to an envelope curve of the Mohrs circles at the failure
moment, where each Mohrs circle is obtained for a different hydrostatic pressure state.
When the envelope is a straight line, (see Figure 9.14), the Mohr-Coulomb criterion
becomes:
W c V N tgG
(9.45)
where W is the shear stress magnitude in the failure plane, V N is the normal stress in the
failure plane, c is the cohesion (material property) and G is the angle of internal friction
(material property). In the particular case when G 0 we revert to the Tresca yield
criterion, with W c , and where the cohesion is interpreted as c k T .
9 PLASTICITY
W c V N tgG
479
W
c - cohesion
V II
V III
VI
V III
V II
VI
G
VN
Figure 9.14: Varying the hydrostatic pressure so as to define the Mohr-Coulomb criterion.
Mohr proved by means of a graph that the equation in (9.45) represents a straight line
which is tangent to the greatest circle defined by V I and V III , (see Figure 9.15). Moreover,
we can also observe that this criterion is independent of the principal stress V II .
W
plastification zone
W c V N tgG
G
V III
V II
VI
VN
V I V III V I V III
cos(2B )
2
2
WA
V I V III
sin( 2B)
2
(9.46)
sin(2B ) c I
cos(2B ) tgG
2
2
2
S
S
V III sin( G) 2c V I V III tgG V I V III cos( G)tgG
2
2
sin G
sin G
V I V III cos G 2c V I V III
V I V III sin G
cos G
cos G
V I
V I
V I
(9.47)
480
S
G
2
S 2B
S
G.
2
VS { W
W c V tgG
R
C
WA
V I V III
2
V I V III
2
R
2B
V III
VI
VN
V NA
C
F ( , c, G) V I
Mohr-Coulomb yield
surface
(9.48)
Then, if we consider the values of V I and V III given in (9.35), the equation in (9.47)
becomes:
J2
1
3
J2
1
3
c cos G
(9.49)
sin S3 R and
J 2 sin S3 R V m sin G
J2
2J 2 and p
(9.50)
6 c cos G
(9.51)
F ( p, q , R , G, c )
q 3 sin R
S
3
2 p sin G q cosR
S
3
sin G
6 c cos G 0
(9.52)
From the above we can verify that for a constant angle R the equation is linear with q and
p . Then, when q 0 we have:
2 p sin G
6 c cos G
6 cos G
c
2 sin G
3 c cotgG
(9.53)
9 PLASTICITY
481
3
1
q 3
2 p sin G q sin G
2
2
6 c cos G
6 c cos G
S
3
(9.54)
6 c cos G
When R
S
3
3
1
q sin G
2
2
6 c cos G
2 6 c cos G
3 sin G
(9.55)
q 3
6 c cos G
(9.56)
6 c cos G
2 6 c cos G
3 sin G
(9.57)
In this way we can draw a graph p q , i.e. to show how pressure varies with the deviatoric
part, (see Figure 9.17).
R 0
q
q
2 6 c cos G
3 sin G
3 c cotgG
S
3
2 6 c cos G
3 sin G
wF
. Then, if we consider the expression of
w
wF
1 sin G and
1 sin G , we can obtain:
wV III
wV I
482
wF
wV ij
n ij
0
1 sin G 0
0
0
0
0 1 sin G
n
n
n ij
1 sin G
2
2 1 sin G
wF
wV ij
1 sin G 2 1 sin G 2
n :n
wF
wV ij
(9.58)
0
0
0
2 1 sin 2 G , with
0
1 sin G
2 1 sin 2 G
(9.59)
Vc1
V1
G
3 c cotgG
V3
Vc3
Vc2
V2
2 k (von Mises)
2 6 c cos G
3 sin G
R
p
p
3 c cotgG
9 PLASTICITY
2 k p tgR
483
q 2 k p tgR
(9.60)
where:
2 6 c cos G
3 sin G
tgR
3 c cotg
2 2 sin G
(3 sin G)
2 sin G
3 (3 sin G)
B 6
(9.61)
From the graph with the p q axes described in Figure 9.19 we can still obtain the
following equations:
2 6 c cos G
3 sin G
2k
2 3 c cos G
3 sin G
6 c cos G
2 k 3 sin G
3 3 sin G
(9.62)
2 6 cos G
Then, by substituting c given above into the equation of p in Figure 9.19 we can obtain:
p
3c
cos G
sin G
2 k 3 sin G cos G
2 6 cos G sin G
3 3 sin G k
2 sin G
3
(9.63)
3 B
Then, with the above results we can represent the Drucker-Prager yield criterion as:
F ( p, q , k , B )
q 2k B 6 p 0
2J 2
with
3 I
3
3 W oct , p
F ( J 2 , I , k, G)
J2 k B I
2 sin G
(9.64)
3 (3 sin G)
2J 2 2 k B 6
3 I
3
0 , thus:
(9.65)
We can still express the Drucker-Prager yield surface in terms of the octahedral stresses, i.e.
by means of V oct
I
(normal octahedral stress) and W oct
3
2
J 2 (tangential octahedral
3
stress), (see Appendix A). Thus, the equation in (9.65) can be rewritten as:
F (V oct , W oct , k, G)
3
W oct k 3B V oct
2
(9.66)
F ( J 2 , I , G ) BI
where B
2 sin G
3 3 sin G
, C
6c cos G
3 3 sin G
J2 C
, I
(9.67)
484
V1
V1
DruckerPrager
MohrCoulomb
V3
V2
V3
V2
3 c cot G
F ( p, q , k , B ) q 2 k B 6 p
F ( J 2 , I , k, B) J 2 k B I
3
W oct k 3B V oct
2
F (V oct , W oct , k, B)
F (J 2 , I , G) BI
J2 C
(9.68)
wF
give us the yield surface gradient in the stress space, so, for
w
the Drucker-Prager yield surface, F ( J 2 , I , k , B )
J 2 k B I 0 , n becomes:
The definition n { F {
wF
w
1
2 J2
s B1
(9.69)
w J2
2 J2
s and
wI
w
1.
Note that according to Figure 9.21 and by using the equations obtained previously,
tgR B 6 , q
2J 2
3 W oct , p
tgR
q q
p
3V m
3 I
3
2 k 3 W oct
3 V oct
(9.70)
whose material parameters are R and k . Then, given any stress state, if the point is on the
line, then the material begins plastification. We can now use the definition in (9.70) to find
the norm (measurement of distance in the stress space):
q ( )
3 KV oct W oct
(9.71)
9 PLASTICITY
485
2k
R
p
B 3
Test 1D
1
Tr ( )
3
D
V1oct
1
f0
3 1D
2
J2
3
D
W1oct
2
f0
3 1D
(9.72)
Test 2D
1
f0
3 2D
2
Tr ( )
3
2D
V oct
2D
W oct
2
f0
2D
3
(9.73)
D
2 k 3 W1oct
2D
2 k 3 W oct
D
V1oct
(9.74)
2D
3 V oct
D
2 k 3 W1oct
D
3 V1oct
2D
2 k 3 W oct
1D 2 D
2 D 1D
2 W oct V oct W oct V oct
2D
D
3
V oct V1oct
(9.75)
and by substituting the equations (9.72) and (9.73) into the above we obtain:
k
f 0 f 0
1D
2D
(9.76)
3 f 01D 2 f 0 2 D
D
2 k 3 W1oct
D
3 V1oct
2 f 01D f 0 2 D
f 01D
3 f 0 2 f 0
2D
1D
1
f0
3 1D
(9.77)
486
f 0 f 0
1D
2D
f 0 f 0 2 f 0 f 0
1D
01D
2
1D
f 0
2D
1D
2D
01D
f 0 f 0
1D
2D
(9.78)
f 0 2 f 0
1D
2D
f 0
2D
(9.79)
f 0
1D
2k 3
2 f 01D f 02 D
f 01D
3 f 0 2 f 0
2D
1D
1
f0
3 1D
D
W1oct
D
3 V1oct
f 0 f 0
1D
2D
f 0 f 0 2 f 0 f 0
1D
01D
1D
2 f 0
2D
1D
1 R0
1 2 R0
9.2.4.3
2D
01D
(9.80)
The Rankine yield criterion also known as the maximum-tensile-stress criterion, (see Chen&Han
(1988)), was formulated by Rankine in 1876. This criterion established that material fails
when the maximum principal stress, V I , reaches the critical value V Tmax . Mathematically,
this criterion is represented by:
VI
V Tmax
(9.81)
where V I ! V II ! V III and V Tmax ! 0 are satisfied. As a result, this criterion is used for
materials that fail only due to traction ( V I ! 0 ). Thus, we can show the Rankine yield
surface as follows:
F (V I , V Tmax )
V I V Tmax
(9.82)
If we now consider the three principal stresses V1 , V 2 , V 3 , the Rankine yield surface is
represented in the principal stress space as shown in Figure 9.22.
Then, by substituting the value of the principal stress V I given by the equation (9.35) into
the Rankine criterion equation (9.82) we obtain:
V I V Tmax
Vm
2
3
J 2 cos R V Tmax
I
2
J 2 cos R V Tmax
3
3
(9.83)
Thus,
F (I , J 2 , R)
I 2 3 J 2 cos R 3V Tmax
(9.84)
2J 2 and p
3V m , the yield surface can now be
Then, if we consider the equations q
rewritten in terms of the Haigh-Westergaard coordinates, i.e.:
F ( p, q , R )
p 2 q cos R 3 V Tmax
(9.85)
Note that the yield curve shape on the deviatoric plane 3 (Nadais plane) can be obtained
by using the equation in (9.31) with p 0 , thus:
9 PLASTICITY
2 q cos R 3 V Tmax
487
3 V Tmax
(9.86)
2 cos R
3 V Tmax
Note that the projection of the deviatoric vector q onto the axis Vc1 is q cos R
which is a constant value. When R 0 q
V1
3 V Tmax
2
and R 60 q
2
6 V Tmax .
V1
V Tmax
V Tmax
V3
V2
V Tmax
V Tmax
V Tmax
V3
V Tmax
V2
3 -plane
V1
V3
V2
V2
V3
488
Vc1
3
(R
60 ) o q
VTmax
(R
0 ) o q
3 VTmax
2
Vc3
Vc2
3 V Tmax
3 V Tmax
(9.87)
2 cos R
3 V Tmax
6 V Tmax .
Then, in summary, the different ways of expressing the Rankine yield surface are:
F (V I , V Tmax )
F ( I , J 2 , R)
F ( p, q, R )
V I V Tmax
I 2 3 J 2 cos R 3V Tmax
p 2 q cos R 3
F (V I , V Tmax )
V I V Tmax
V Tmax
(9.89)
wF
, the gradient of the Rankine yield surface given
w
n ij
wF
wV ij
wV I
wV ij
1 0 0
0 0 0
0 0 0
(9.90)
9 PLASTICITY
489
V1
octahedral plane
I1
V2
0 ( 3 -plane)
VI
V II
VI
V II
V III
S
6
pure shear
S
6
V3
V III
R 0
3 max
VT
2
IT
p0
3V Tmax
IC
6 VTmax
R 60
9.2.5
In order to fully describe material behavior, we must not just look at the onset of
plastification, but rather at its evolution too, i.e. at how such material behavior evolves
when undergoing loading/unloading/loading. From a material point of view, the yield
surface can be altered after the start of plastification, but such changes depend on the type
of material being analyzed. Next, we will discuss some ways of how the yield surface
evolves, Chen&Han (1988).
The simplest model to characterize material behavior during plastification is the so-called
elastic-perfectly plasticity model, which is characterized by a uniaxial stress-strain curve (as
shown in Figure 9.28). In this scenario the yield surface remains unchanged during
plastification. That is, the yield stress value is not affected, and the yield surface remains
unaltered as plastification evolves.
490
Another idealized model used to show how plasticity evolves is known as hardening
plasticity where we can emphasize two basic models: the isotropic and the kinematic. In
uniaxial cases, the isotropic hardening plasticity model is represented in Figure 9.32. As we can
see in said figure, as plastification evolves the elastic range develops symmetrically. In
three-dimensional cases and (if we dealing with the Drucker-Prager yield surface) the
evolution of this yield surface is as shown in Figure 9.26 in which we can observe that it
does not change its shape but, rather, expands symmetrically during plastification.
V1
V1
V2
V3
Vc1
V3
Vc2
V2
Figure 9.26: Evolution of the yield surface in the Drucker-Prager model with isotropic
hardening.
The kinematic hardening plasticity model is characterized by the fact that the size and the shape
of the elastic range do not change whilst plasticity evolves, but the elastic range is able to
move. In Figure 9.37 we show a uniaxial stress-strain curve whose yield surface evolution
neither changes its shape nor expands, but rather changes its position. Then, in a
bidimensional case example, in which the initial yield curve is represented by the
circumference in the principal stress space, if we dealing with the kinematic hardening
model the yield curve only moves in this space, (see Figure 9.27).
V2
V1
9 PLASTICITY
491
It is possible to formulate more complex models by combining basic models, such as the
isotropic-kinematic hardening plasticity model which considers isotropic and the kinematic
behaviors simultaneously, i.e. the yield surface expands and can also move.
Then, we summarize some criterion to describe how plasticity evolves:
x Perfect Platicity
( The surface does not evolve)
Yield Surface
x Kinematic Hardening Plasticity
( The surface does not change its shape, just moves)
x Isotropic - Kinematic Hardening Plasticity
( The surface expands and moves)
The following section will deal with the mathematical formulations that govern the models
described above. For simplicity, we will use the one-dimensional case to describe and
formulate the mathematical expressions that characterize each of these models, and then
we will extend these models to the three-dimensional case (3D), (see Simo&Hughes
(1998)).
Next, a few one-dimensional plasticity models will be described. However in some cases
these simple models do not describe how most materials really act. Nevertheless, they are
useful to gain and understanding of the plasticity mechanism, and can also be used to
establish more complex models.
9.3.1.1
called the plastic strain denoted by H p , and the corresponding recoverable strain is called
(1)
the elastic strain, H e .
The rheological model (a device used to interpret physical phenomena) used to represent
the perfect elastoplastic behavior is made up of a linear spring and a Coulomb friction
device in series, (see Figure 9.30). The behavior of each device in isolation can be
appreciated in Figure 9.29. Note that until stress V Y is reached, the friction device does not
suffer permanent deformation, and when it reaches the stress limit (which is not permitted
to exceed V Y ), it starts to deform without increasing stress. Note also that, when subjected
to unloading, it maintains the strain value at the start.
492
VY
5
1
Hp
4
(1)
He
Hp
E
1
(1)
He
(2)
(2)
VY
loading
unloading/loading
E
VY
Hp
He
(1)
Hp
VY
VY
E
VY
9 PLASTICITY
493
there being any increase in stress (see Figure 9.28, steps 2-3-6). If there is any unloading the
strain undergone by the Coulomb friction device can not be recovered, and the elastic
device starts to recover its elastic strain, (see Figure 9.28 at the branch 4 or 7).
Next, we will establish the mathematical model that characterizes this type of behavior.
We can verify that the following holds:
He
EH e
E H Hp
H Hp
(9.91)
(9.92)
If by means of the material behavior described in Figure 9.28, we can check that the
magnitude can never be greater than V Y , then we can define the admissible stress space,
E , such as:
^V R
V V Y d 0`
F (V )
F (V )
(9.93)
0 , then:
V VY 0
H p
(9.94)
We can define the interior of that admissible stress space ( int E ) as:
int E
^V R
F ( V)
V V Y 0`
(9.95)
^V R
wE
F (V )
V VY
0`
(9.96)
wE int E
wE int E
(9.97)
For one-dimensional case, the yield surface is limited by two points, and does not change
during a loading/unloading/loading process, (see Figure 9.31).
Note that, since the stress space does not support the stress value V ! V Y thus F (V) ! 0 ,
and also when F (V)
V V Y 0 it implies that H p
We can now analyze what happens when the stress has negative values ( V 0 ). To do so,
we will adopt the scalar H t 0 known as the plastic multiplier. Then, it follows that:
H p
H
H t 0
iff
H d 0
V VY ! 0
iff
V V Y 0
(9.98)
H sign(V)
(9.99)
494
if
V!0
if
V0
Sign function
(9.100)
V
VY
wE
VY
H
w'
wV ij
(9.101)
where ' is a potential. In the particular case when ' is equal to the yield surface, i.e.
' F , the flow rule is said to be associated, which is what happens in the case under
consideration:
H p
H
wF
wV
H
w V VY
wV
H
w V
wV
H sign(V)
(9.102)
(9.103)
H t 0
if
if
F ( V) d 0
F ( V)
F ( V)
V VY 0
H 0
V VY
H z 0
Now, all the conditions above can be unified into one single condition H F (V) 0 . Then,
we can introduce the Kuhn-Tucker conditions:
H t 0
F ( V) d 0
H F (V) 0
(9.104)
Notice that, with respect to H , we have not yet fully characterized the mathematical model.
Let us now consider a general scenario in which: at a certain instant in time t the variables
H(t ) , H p (t ) and V(t ) are known, and F (V) 0 is fulfilled, and by the fact that the point is
in the plastification process we have H ! 0 . The rate of change of F (V) can now be
approached by time discretization by means of the intervals 't in such a way that
'tF (V) Ft 't (V) Ft (V) 0 is satisfied. Note that, for F (V) ! 0 it implies that
Ft 't (V) ! 0 , whose result is inadmissible. That is, for any given stress state, if the point is
9 PLASTICITY
495
outside the yield surface, the only way in which the condition Ft 't (V) 0 is satisfied
occurs if the yield surface evolves, F (V) 0 . In perfect plasticity, this point cannot be
outside the yield surface, the only possibility is by moving on the yield surface. Then,
mathematically, we can summarize the previous comment as follows:
H ! 0
iff
H 0
iff
F 0
F 0
H F 0
(9.105)
(9.106)
Then, we can summarize perfect elastoplastic behavior, (see Simo&Hughes (1998)), by:
i.
ii.
Flow rule
H p H sign(V)
The perfect
(9.107)
elastoplastic model
9.3.1.2
Consistency condition
H F 0
Let us suppose now that a material point (particle) in a hypothetical material is undergoing
the action of loading/unloading/loading in such a way that the stress-strain curve is that
indicated in Figure 9.32. We can describe this behavior as Isotropic Hardening Elastoplasticity.
With regards to this figure, the loading and unloading steps are indicated by the numbers:
1, 2, 3, 4, 5, 6. We will check, once again, that after unloading step 4 there is a residual
(1)
(1)
strain H p (plastic strain) and a corresponding recoverable strain H e (elastic strain). The
difference between this and perfect elastoplastic behavior is: as the plastic strain evolves the
elastic limits does so also, but symmetrically, i.e. the yield surface evolves symmetrically. As
seen in Figure 9.32, before plastification begins the elastic limit in the material is defined by
>VY ,V Y @ . When the stress state reaches point 3 unloading is applied, and we can then
verify we have a new elastic limit defined by [V Y* ,V *Y ] . Then, because the yield surface
expands symmetrically this model is known as the isotropic hardening elastoplastic model.
The rheological model that shows isotropic hardening elastoplastic behavior is made up of
a linear spring, a Coulomb friction device in parallel, and another linear spring device in
series as shown in Figure 9.33.
496
V *Y
2
VY
K eq
5
E
1
Hp
E
4
(1)
He
(1)
VY
V *Y
VY
Hp
He
Figure 9.33: The rheological model for the isotropic hardening elastoplastic model.
Physically speaking, the rheological model shown in Figure 9.33 can be interpreted as
follows. Initially, we just have the elastic strain, H e , which is caused by the spring with the
constant E (since the Coulomb friction device does not deform for stress values less than
V Y and because of this the spring with the constant K does not undergo deformation
either). This load stage is typical of the initial elastic domain, and is shown by the branch 1
in Figure 9.32. Note that here V EH e V EH e holds, i.e. the secant modulus and the
tangent modulus are the same, i.e. E S E T E . Then, when the stress level reaches the
value V Y , the Coulomb friction device starts to deform almost freely, i.e. the Coulomb
friction device strain is controlled by the spring of constant K , and at this stage all the
stress is absorbed by the springs. This stage is represented by the branch 2 3 in Figure
9.32. Note that, in this model, the parameter K is constant during the deformation
9 PLASTICITY
497
process, so, V V Y KH p V KH p holds. Here, with 2 3 , we can state that the
equivalent spring is that which results from those two in series, as shown in Figure 9.34.
The constant of the equivalent spring, K eq , is the tangent of the branch slope 2 3 6 .
H H p He
V
H H p H e
V V ( K E )
H
V
K E
KE
KE
V
H
(K E )
V K eq H
H p
H e
V
K eq
V
V
where
K eq
KE
(K E )
H
stress (unloading) occurs, the plastic strain in the Coulomb friction device, H p , is
maintained and consequently the strain in the spring K is also, which causes a new elastic
(1)
He
H Hp
EH e
E H Hp
of change
rate
o V
(9.108)
E H H p
(9.109)
For isotropic hardening elastoplastic behavior the yield criterion is given by:
F (V, KB)
V V Y KB d 0
(9.110)
where V Y ! 0 and K ! 0 are material constants and where K is known as the plastic
modulus, (see Simo&Hughes(1998)). Note that the term KB has unit of stress, K has the
same unit as E (unit of stress), and B t 0 plays the role of plastic strain which is
dimensionless, (see Figure 9.35). Then, by adopting the hardening hypothesis B H p and
by considering that H p
We can define the admissible stress space, (see Figure 9.35(a)), as:
498
^V R
V V Y KB d 0`
F (V, KB)
(9.111)
^V R
int E
F (V, KB)
V V Y KB 0`
(9.112)
^V R
wE
F (V, KB)
V V Y KB 0`
(9.113)
The elastic domain, int E , together with its boundary, wE , describes the admissible
stress space E , i.e.:
E int E wE
V v H p
V
VY
K
1
VY
wE
K - Plastic modulus
V *Y
KH p
V *Y
(9.114)
VY
Hp
V *Y
a) Stress space
Figure 9.35: Isotropic hardening elastoplastic behavior.
The Kuhn-Tucker conditions are still valid:
H t 0
F (V, KB) d 0
H F (V, KB) 0
(9.115)
(9.116)
The consistency condition allows us to obtain the explicit relationship for H . That is, when
H ! 0 , we must satisfy F F (V, KB ) 0 :
F (V, KB)
wF D( KB )
wF
V
wKB Dt
wV
wF
) 0
V (1)( K B KB
wV
(9.117)
wF
sign(V) . In addition, if
wV
H , the
0 , and by considering V E H H p and B
9 PLASTICITY
F (V, KB)
499
(9.118)
F (V, B)
sign(V) E H E H K H
sign(V) E H
E K
H
(9.119)
H sign(V)
H p
E H
E K
(9.120)
E H H p
E H
E H
E K
EK
H
E K
E ep H
(9.121)
where E ep is the elastoplastic tangent stiffness modulus which is the same as the constant of the
equivalent spring obtained in Figure 9.34.
V
VY
E ep H
V
E ep
E ep
E
E
EK
E K
E H
E K
E K H p
H
(9.122)
EK
H
E K
E K E K p
H
E K E
V
K H p
(9.123)
Note that K is the line slope of the graph V u H p described in Figure 9.35(b).
The constitutive relations are now defined as follows:
F (V, KB) 0
V int E `
V
E H
(9.124)
V
E H
(9.125)
F (V, KB) 0
V int E
500
F (V, KB)
V wE
EK
V
E K
H
E ep H
(9.126)
ii.
Flow rule
H p H sign(V)
The isotropic
hardening
elastoplastic
model
B H
iv. Yield condition
F (V, KB)
v.
V V Y KB d 0
(9.127)
9.3.1.3
Kinematic hardening elastoplastic behavior was first described by Prager (1955). In this
model the shape of the yield surface, in the principal stress space, is not altered, however it
can move. In uniaxial cases the yield surface is made up of two points, and additionally, in
kinematic hardening the distance between these does not change during plastification, (see
Figure 9.37).
Now, the yield criterion for kinematic hardening elastoplastic behavior is given by:
F (V, q)
V q VY
(9.128)
The internal variable q is associated with the new position of the yield surface center in the
principal stress space, (see Figure 9.37) as well as being a function of the plastic strain, for
example we can adopt a linear relationship, e.g.:
q
H H p
(9.129)
where H is known as the kinematic hardening modulus (material property) and generally
speaking it is a tangent modulus of the curve q u H p , (see Figure 9.37). In the case under
consideration we will consider H to be a constant variable.
In this model the rate of change of the plastic strain is defined as follows:
H p
H p
H ! 0 if
H 0 if
V Y 0
V q VY ! 0
Vq
H p
H sign(V q)
(9.130)
where
1
sign(V q )
1
if
Vq!0
if
Vq0
(9.131)
H H p
H H sign(V q )
(9.132)
9 PLASTICITY
501
V *Y
VY
H
o*
E ep
5
q
o
Hp
Hp
E
1
4
(1)
He
(1)
V *Y
VY
ii.
Flow rule
H p H sign(V q )
The kinematic
hardening
elastoplastic
model
(9.133)
F (V, q)
wF
wF
V
q
wq
wV
(9.134)
wF w V q
w Vq
wV
sign(V q) ,
wF
wq
wF w V q
w Vq
wq
502
wF
wF
q 0
V
wV
wq
sign(V q ) V sign(V q) q 0
F (V, q)
(9.135)
Thus
H
sign(V q) E
H
>E H @
H
>E H @ H
sign(V q) E
(9.136)
H p
sign(V q ) E
H sign(V q )
>E H @
>E H @
H
(9.137)
Then, by means of the rate of change of the stress-strain relationship we can obtain the
elastoplastic tangent stiffness modulus E ep , i.e.:
V
E H H p
E
H
E H
>E H @
EH
>E H @
H
E ep H
V
>E H @
EH
E ep
>E H @
(9.138)
H ! 0
(9.139)
9.3.1.4
E H
if
H 0
and
EH
H
if
This model is a combination of the models discussed previously. Here, the yield criterion
can be defined as follows:
F (V, q, KB)
V q V Y KB d 0
(9.140)
ii.
Flow rule
H p H sign(V q )
B H
V q V Y KB d 0
The isotropickinematic
elastoplastic
model
(9.141)
9 PLASTICITY
503
Once again, we will use the consistency condition, F H 0 , to obtain H . Firstly, we will
calculate the rate of change of F (V, q, B ) by means of the chain rule of the derivative, i.e.:
F (V, q, KB)
wF D( KB )
wF
wF
d0
V
q
w ( KB ) Dt
wq
wV
(9.142)
wF w V q
w Vq
wV
sign(V q) ,
wF
wq
wF w V q
w Vq
wq
sign(V q) ,
wF D( KB )
, and also if we consider that V E H H p , B
KB
w ( KB ) Dt
q H H sign(V q ) , the equation in (9.142) becomes:
H and
wF
wF D ( KB )
wF
0
V
q
wV
wq
w ( KB ) Dt
0
sign(V q) V sign(V q) q K B
F (V, q, KB)
(9.143)
Thus
H
sign(V q) E
H
>E H K @
H
>E H K @ H
(9.144)
sign(V q) E
after which we can obtain the rate of change of the plastic strain as follows:
H p
H sign(V q )
sign(V q) E
H sign(V q)
>E H K @
>E H K @
H
(9.145)
Then, by means of the rate of change of the stress-strain relationship we can obtain the
elastoplastic tangent stiffness modulus E ep , i.e.:
V
E H H p
E H
H
>
@
E
H
K
E H K
H
>E H K @
E ep H
E ep
E H K
>E H K @
(9.146)
E H
if
H 0
and
V
E H K
H
>E H K @
E ep H
if
H ! 0
(9.147)
504
There is material isotropy. That is, during plastification the material does not lose
its isotropy;
The process is isothermal and adiabatic. That is, during plastification, the effect of
temperature is not taken into account, i.e. the constitutive equation is independent
of temperature.
The production of plastic strain is associated with internal energy dissipation, i.e. with
irreversible processes. To put it another way: a process which involves plastic strain is path
dependent, for example, in Figure 9.38 the points A and B are associated with the same
stress but have different strain values, so, here we have a process that depends on the load
history. Now, we will adopt the constitutive equation with internal variables, (see
subsection 6.4.1 in Chapter 6), in order to describe the motion history, albeit indirectly.
V
VA
VB
H (B )
H ( A)
Hp
He
9.4.1
With small deformation, we use additive decomposition of the infinitesimal strain tensor
into elastic and plastic parts, i.e.:
Additive decomposition of the
infinitesimal strain tensor
e p
(9.148)
where is the infinitesimal strain tensor, whose components in the Cartesian basis are
H ij H eij H ijp , (see Chapter 2 in subsection 2.14). The assumption in (9.148) forms the basis
of the classical theory of plasticity, which is valid in a small deformation regime.
Then, if we consider (9.148) the constitutive equation for stress becomes:
Ce : e
Ce : p
Ce
1
(9.149)
where C e M1 1 2NI is the elasticity tensor for isotropic materials, and the elastic
1
M
1
compliance tensor is given by C e
1 1
I , where M , N are the Lam
2N(3M 2N)
2N
constants, (see Chapter 7).
9 PLASTICITY
9.4.2
505
We apply the constitutive equation with internal variables, (see Chapter 6), where, in
general, the Helmholtz free energy, Z Z ( , T , Bk ) , is described in terms of the
infinitesimal strain tensor ( ), temperature ( T ), and a set of internal variables ( Bk ), which
could be a scalar, vector, or higher order tensor. Then, if we consider the process to be
isothermal, we have Z Z ( , Bk ) . We can now reformulate this energy expression, in
order to obtain the elastic part of as a free variable, i.e.: Z Z ( e , B k ) . Now the set of
internal variables B k do not include the plastic component ( p ) of the strain tensor,
because it is already included in the free variable e p . Then, the following is
satisfied:
wZ
w e
wZ wZ
w w p
(9.150)
wZ e
wZ
k
:
B
wB k
w e
(9.151)
9.4.3
1 q& T t 0
: D SIT SZ
x
T
isothermal
process
o
Dint
t0
: D SZ
(9.152)
Note that, in isothermal processes the internal energy dissipation is purely mechanical. By
substituting the rate of change of the Helmholtz free energy given in (9.151) into the
Clausius-Duhem inequality, and also if we know that D E | e p holds in a small
deformation regime, we can obtain:
wZ
wZ
kt0
B
: ( e p ) S e : e
B
w
w
k
wZ e
p
k t 0
Dint S e : : Ak B
w
Dint
S
wZ
wB k
(9.153)
(9.154)
inequality in (9.154) must be valid for any admissible thermodynamic process, (see Chapter
k 0 , then the following must hold:
6), so, we can adopt a process in which p 0 and B
506
wZ
w e
(9.155)
Note that is related to the gradient of Z in the strain space, and Ak is related to the
gradient of Z in the space of B k .
Then, by considering the constitutive equation in (9.155), the internal energy dissipation
(9.154) becomes:
k t0
: p Ak B
Dint
(9.156)
Next, we apply the maximum dissipation principle, which states that the dissipation in a
material reaches a maximum during a change characterized by a dissipative process. Let us
consider the current state ( , Ak ) , which is the current distribution of Cauchy stress tensor
and thermodynamic forces in a body subjected to plastic strain. The principle of maximum
energy dissipation requires that for a change of state, represented by ( * , Ak * ) , the
following must be satisfied:
k t0
( * ) : p ( Ak Ak* )B
(9.157)
The inequality in (9.157) describes an optimization problem with constraint. We can also
maximize the dissipation by minimizing the negative dissipation with the constraint
'( , Ak ) d 0 . To this end, we define the Lagrangian as:
k H '
: p Ak B
L Dint H '
(9.158)
p H
w' ( , Ak )
w
p
H
w' ( , Ak )
w
(9.159)
k H
B
w'
wAk
k
B
H
w'
wAk
(9.160)
H
k
B
H
wG
w
wG
wAk
(9.161)
(9.162)
Then to fully define the model we need to introduce the loading/unloading Kuhn-Tucker
conditions:
H t 0
F ( , Ak ) d 0
H F ( , Ak ) 0
(9.163)
(9.164)
9 PLASTICITY
507
In associated flow, the plastic potential G is equal to the yield surface F , i.e.
G F Associated flow
(9.165)
We can now summarize the elastoplastic model for isothermal processes under the small
deformation regime as:
i.
w e
iii. Plastic flow rule
wG
p H
w
iv. Evolution of the internal variables
wG
ii.
k
B
v.
H
(9.166)
wAk
Kuhn - Tucker complementary condition
; F ( , Ak ) d 0
vi. Consistency condition
H t 0
H F ( , Ak ) 0
H F ( , Ak ) 0
9.4.4
In this section we will establish the rate of change of the stress-strain relationship, i.e.:
(9.167)
C tan_ep :
tan_ep
Ce : e
where C e is the elasticity tensor, which does not vary with time ( C e O ), and e is the
elastic part of the infinitesimal strain tensor. Then, by means of the strain tensor additive
decomposition, (see Eq. (9.148)), we can express the elastic part as e p , and by
substituting this into the equation in (9.168) we obtain:
of change
C e : ( p ) rate
o
Ce : e
H
C e : ( p )
wG
w
wG
C e : ( p ) C e : H
w
(9.169)
F ( , Ak )
wF
:
w
wF
Aa
wA
a 1
(9.171)
508
A a
wAa
: p
w p
H
wH
wAa
H
w p
wH
wAa
H
w p
wH
(9.172)
F ( , Ak )
wF
Aa
wA
a 1
n
wF
wF
: H
w
w
a 1 Aa
wAa w p
w p : wH 0
(9.173)
Next, by substituting the equation in (9.170) into the one above we obtain:
n
wG
wF
: H H a :
w
w
a 1
F ( , Ak )
wF e
: C
w
n
wG
wF
H
: H
w
a 1 wAa
wAa w p
w p : wH 0
n
wG
wF wAa w p
wF
wF
H
: C e : H
: Ce :
p : wH
w
w
w
a 1 wAa w
(9.174)
wF
: C e :
w
n
wG
wF
wF wAa w p
: Ce :
:
w
w a 1 wAa w p wH
(9.175)
Then, drawing once more on the equation in (9.170) and if we consider the plastic
multiplier obtained in (9.175), it follows that:
wG
Ce
C e : H
w
wG
: H
w
wF
: C e :
w
G
e
w
C :
n
p
wF
wF wAa w w
e wG
:C :
:
w
w a 1 wAa w p wH
(9.176)
wF
,m
w
wG
w
, where n
and m are symmetric second-order tensors after which the equation in (9.176) can be
rewritten as:
e
C
:
:
e
C :
m
n
p
wF wAa w
n : Ce :m
p : wH
a 1 wAa w
(9.177)
Note that the denominator is a scalar, and by denoting the inverse of the denominator by
X we obtain:
C e : X C e : n : C e : m
(9.178)
9 PLASTICITY
e
e
C ijkl
H kl X C ijkl
n ab C eabcd H cd m kl
V ij
e
C ijkl
H st
C
C
e
C ijkl
E sk E tl X
E sk E tl X
e
C ijkl
509
n ab C eabcd H st
e
C ijkl
n ab C eabcd
e
ijst
X
e
C ijkl
n ab C eabst m kl
e
ijst
X
e
C ijkl
m kl n ab C eabst
H
H
E sc E td m kl
E sc E td m kl H st
(9.179)
st
st
e
e
e
C :m n : C
:
C
n
p
wF wAa w
n : C :m
p : wH
a 1 wAa w
X C e : m n : C e :
(9.180)
C tan_ep :
(9.181)
tan_ep
wG
wF
w
C
n
w
G
wF
w
F
: Ce :
w
w a 1 wAa
Ce :
: Ce
wAa w p
w p : wH
(9.182)
Note that C tan_ep is a fourth-order (but, not necessarily symmetric) tensor, showing only
minor symmetry due to that and . Then, in terms of n and m the elastoplastic
tangent stiffness tensor becomes:
C tan_ep
Ce
Ce :m n : Ce
n
wF wAa w p
n: Ce :m
p : wH
a 1 wAa w
(9.183)
Ce
Ce :n n : Ce
n
wF wAa w p
n: Ce :n
p : wH
a 1 wAa w
(9.184)
In associated flow, the elastoplastic tangent stiffness tensor, C tan_ep , is a symmetric tensor,
i.e. it features both major and minor symmetry.
Problem 9.1: Consider a one-dimensional case, find the elastoplastic tangent stiffness
tensor (elastoplastic tangent stiffness modulus) for the following: 1) perfect plasticity; 2)
isotropic hardening plasticity, 3) kinematic hardening plasticity, and 4) isotropic-kinematic
hardening plasticity.
Solution:
We start from the general definition of the elastoplastic tangent stiffness tensor given in
(9.182), in which the associated flow G F takes place. Then:
510
tan_ep
wF
wF
w
w
Ce
n
wF
wF
e wF
:C :
w
w a 1 wAa
Ce :
1) Perfect plasticity
With perfect plasticity, the yield surface is given by:
F (V )
V VY
wF
wV
: Ce
wAa w p
w p : wH
sign(V)
H sign(V)
n
wF wAa w p
p : wH
a w
wA
a 1
0.
Furthermore, as we are dealing with a one-dimensional case, the tensors can be represented
tan_ep
by their only nonzero component, i.e. V 11 C1111
H 11 V E tan_ep H , and
C tan_ep o E tan_ep
Thus, as expected:
E tan_ep
E
Ce o E
wF
wF
o
wV
w
>Esign(V)@ >sign(V) E @
EE
sign(V) Esign(V) 0
V V Y KB
wF
wV
wF
wA
sign(V)
wF
w ( KB )
1
H sign(V)
wF wAa w p
p : wH
a w
a 1
1
wF wA1 w p
p : wH
a 1 wA1 w
The term
wA
B H
is evaluated as follows:
1
wF wKB wH p
o
p wH
a 1 wKB wH
wH p
wH
wB
wH p
sign(V)
sign(V)
1Ksign(V)sign(V) K
E
Esign(V)sign(V) E
sign(V) Esign(V) ( K )
E
E2
EK
EK
EK
9 PLASTICITY
F (V, q, )
V q VY
511
wF
wV
wF
wq
sign(V q)
sign(V q)
q H H p
wH p
wH
sign(V)
wF wAa w p
1
1
wF wA1 w p
wF wq wH p
o
:
p
The term
wA
E
Esign(V)sign(V) E
sign(V) Esign(V) ( H )
E
E2
EH
EH
EH
V q V Y KB
wF
sign(V q)
wV
wF
sign(V q )
wq
wF
1
wKB
Thus:
wF wAa wH p wF wA1 wH p wF wA2 wH p
wA
wF wAa wH p
p wH
a wH
wA
a 1
wF
wA1
wA1 wH p wF
wH p wH wA
2
wA2 wH p
wH p wH
wF
wq
E
E2
E H K
E H K
EH K
512
9.4.5
9.4.5.1
Perfect Plasticity
F ( )
13 >Tr ( )@ R
2
with
(9.185)
2
V
3 Y
where R is the radius of the yield surface, and V Y is the yield stress (elastic limit
stress), and is the Frobenius norm of the Cauchy stress tensor, (see subsection
1.5.7 Norms of Tensors in Chapter 1).
3. There is Levy-Saint_Venant associated plastic flow
4. There is no hardening
Note that F ( ) can be expressed as follows:
F ( )
13 >Tr ( )@ R
2
13 Tr( )1 : R
: 13 Tr ( )1 : R
s: R
(9.186)
where s { dev , i.e. it is the deviatoric part of the Cauchy stress tensor. Then by substituting
s 13 Tr ( )1 into the above equation we obtain:
F ( )
s: R
s : s 13 Tr ( )1 R
s :s R
(9.187)
s R
where we have taken into account that the trace of any deviatoric tensor is equal to zero, i.e.
s : 1 Tr (s ) 0 .
We then obtain the gradient of F ( ) in the stress space, the result of which is the plastic
flow tensor n :
n
wF ( )
w
Note that
w
w
> s : s R@
1
1
s : s 2 ws : s s : ws
w
2
w
1
s : s 2
ws
:s
w
(9.188)
ws
: s s , (see Problem 1.39) with which we obtain:
w
n
wF ( )
w
1
s : s 2
ws
:s
w
s
s :s
s
s
(9.189)
n : n 1.
With the above, we can obtain the plastic multiplier such as that defined in (9.175) for the
associated flow case:
H
wF
: C e :
w
n
wG
wF
wF wAa w p
: Ce :
:
w
w a 1 wAa w p wH
n : C e :
n : Ce :n 0
(9.190)
9 PLASTICITY
513
where the elasticity tensor, (see Chapter 7), can be expressed as follows:
Ce
M1 1 2NI N1 1 2N>I 13 1 1@
(9.191)
n : M1 1 2NI :
n : M1 1 2NI : n 0
(9.192)
Thus
H
n : C e :
n : Ce :n 0
Mn : 1 1 2Nn : I :
Mn : 1 1 : n 2Nn : I : n
n :
n:n
n :
s
s
with
(9.193)
We can also find the elastoplastic tangent stiffness tensor for the associated flow case,
given in (9.184):
C tan_ep
Ce :n n : Ce
n
wF wAa w p
n : Ce :n
p : wH
a 1 wAa w
Ce
(9.194)
M1 1 2N>I n n@
N1 1 2N>I 13 1 1 n n@
(9.195)
for when H ! 0 .
9.4.5.2
This model has two internal variables, namely ^A1 q; A2 B ( KB )` where B (scalar) is the
equivalent plastic strain which defines isotropic hardening behavior and q (a second-order
tensor) defines the center of the von Mises yield surface in the deviatoric stress space. In
this model we have the following hardening law and plastic flow rules:
I sq
F ( , A1 , A2 ) I
H
p
Tr (q) 0
B)
2
K(
3
H
I
I
B)
2
H(
3
I
I
(9.196)
;
B H
2
3
where s { dev , K (B) is the isotropic hardening modulus, and H (B ) is the kinematic
hardening modulus. Then, given that p
follows:
t
B (t )
2 p
(W) dW
3
(9.197)
514
K (B ) V Y K B
(9.198)
n : C e :
n
wF wAa
n : Ce :n
p
w
a 1 Aa w
2Nn :
n
wF wAa w p
2N
p : wH
a 1 wAa w
wF wAa w p
p : wH
a w
wA
a 1
2Nn :
p
w
wH
2Nn : n
a 1
I
I
(9.199)
wA1 w p wF wA2 w p
w p : wH wA w p : wH
2
wF wq w p
wF w ( KB ) w p
:
:
wq w p wH w ( KB ) w p
wH
2
I I
sign(I )
:
sign(I ) H (B )
3
I
I
2
2
Hn : n Kn : n
3
3
2
2
H K
3
3
where n
wF wAa w p
p : wH
a w
wA
with
I I
2
2
sign(I )
sign(I )
:
K
I I
3
3
(9.200)
1.
2Nn :
wF wAa w p
p : wH
1 wAa w
2Nn :
2
2
2N H K
3
3
H
n :
HK
1
3N
(9.201)
tan_ep
nn
N1 1 2N I 13 1 1
HK
1
3N
(9.202)
9 PLASTICITY
515
wU c
.
w
Similarly, von Mises suggested there was a plastic potential function, G ( ) , with which the
rate of change of the plastic strain is given by:
p
H
wG
H ijp
H
wG
(9.203)
wV ij
where H is the plastic multiplier- a positive scalar. This theory is known as the plastic
potential theory. One possible approach we can take to this is to consider the plastic potential
to be equal to the yield surface G F , with which it is said that the flow is associated.
Otherwise, i.e. when G z F , there is said to be non-associated flow rule. In the case of the
former we obtain:
p
H
wF
w
H ijp
H
wF
wV ij
(9.204)
The plastic work done by an external agency, during the application of additional
stress, is positive.
The total work done by an external agency during a cycle can not be negative.
We will discuss these rules in the following example. First, let us consider a stress
relationship as shown in Figure 9.39.
V
Stable
Unstable
V (1)
dV
VY
V0
V *(1)
V*
V *(1)
a)
Hp
b)
516
Let us consider the stress state V ij which is inside the initial yield surface at the initial
instant of time ( t 0 ). At time t1 , the point is found on the yield surface, so the process
observed between t 0 and t1 is purely elastic. From t1 to t 1 Gt , we apply a load
increment described here by dV , and then we apply unloading, as shown in Figure 9.40.
initial yield surface
V dV
( t0 )
( t 1 Gt )
( t1 )
t1
dWT
V ij dH eij dt
t*
V ij dH ije
dH ijp
dt V
e
ij dH ij dt
t1 Gt
t1
t1 Gt
V ij dH ije dt
p
ij dH ij dt
t1
t1 Gt
(9.205)
V ij dH ijp dt
t1
t1 Gt
p
*
ij dH ij dt
, and therefore,
t1
t1 Gt
dWT dW0
dWn
t1 Gt
V ij dH ijp dt
t1
t1 Gt
V *ij dH ijp dt
t1
ij
V *ij dH ijp dt
(9.206)
t1
dW n t 0
ij
V *ij dH ijp dt ! 0
(9.207)
t1
which represents Druckers second postulate and because the above integrand is valid at
any time, it holds that:
ij
Then, with dV ij
:
V *ij dH ijp t 0
t0
(9.208)
V ij V *ij we obtain:
dV ij dH ijp t 0
d : d p t 0
(9.209)
Under these conditions the material is said to be plastically stable. For a material with
hardening behavior we obtain:
dV ij dH ijp ! 0
d : d p ! 0
(9.210)
9 PLASTICITY
517
d : d p
(9.211)
Let us now suppose that there is a scalar-valued tensor function F F ( ) called the
plastic potential or yield function. In the elastic regime the following is satisfied:
F 0
dF
wF
dV ij 0
wV ij
(9.212)
dF
wF
dV ij t 0
wV ij
(9.213)
(9.214)
Now, to solve a problem with a restriction we will introduce the Lagrange multiplier dO :
dV ij dH ijp dH
wF
dV ij
wV ij
p
dH ij dH wF
wV ij
dV ij
(9.215)
dH
wF
wV ij
d p
dH
wF
w
(9.216)
where dH is a positive scalar. The equation (9.216) is called Prandtl-Reusss flow rule.
Note that in an isotropic material the yield surface can be expressed in terms of the three
principal stresses, i.e. F F (V1 , V 2 , V 3 ) . In this case, d p can be represented by a vector in
the principal stress space, (see Figure 9.41).
Note also that
wF
{ F
w
is the
V2
dV
dH p
dV
F 0
V1
518
(9.217)
&
&
F e ( X , t) F p ( X , t)
&
&
E e ( X , t) E p ( X , t)
(9.218)
(9.219)
Kinematic Tensors
&
&
F e ( X , t) F p ( X , t)
Multiplicative decomposition
(9.220)
where F e is the elastic transformation, and F p is the plastic transformation, (see Figure
9.42). Then, according to Figure 9.42 the following is satisfied:
&
dx
&
F dX
&
F e F p dX
(9.221)
Note that, first we make the transformation related to F p , thereby defining a new
configuration called the intermediate (or stress-free) configuration in which it holds that
&
dX
&
&
F p dX . Then, we make the transformation associated with F e , where dx
&
F e dX
holds, (see Figure 9.42). Next, from the multiplicative decomposition we can obtain the
following relationships:
F
Fe F p
F 1
Fp
1
Fe
1
F 1 F e
Fp
1
(9.222)
9 PLASTICITY
519
Next we will establish the kinematic variables in the intermediate configuration B , and
show how these are related to those defined in the reference and current configurations.
&
dX
&
dX
B0
&
X
Fe
intermediate
configuration
reference
configuration
&
dx
&
X
current
configuration
&
x
QF p
intermediate
configuration II
F e
&
X
F e QT
QT
&
dX
B
Fp
&
dX
B0
reference
configuration
&
X
&
X
Fe
intermediate configuration I
Fe F p
&
dx
B
&
x
current
configuration
520
9.7.1.1
Now, remember from Chapter 2 that the right Cauchy-Green deformation tensor
&
&
( C ( X , t ) ), the left Cauchy-Green deformation tensor ( b( x, t ) ), the Green-Lagrange strain
&
&
tensor ( E ( X , t ) ), the Cauchy deformation tensor ( c( x, t ) ), the Almansi strain tensor
&
&
&
( e( x, t ) ), the right stretch tensor ( U( X , t ) ), and the left stretch tensor ( V ( x, t ) ) are related
to each other as shown in Figure 9.44.
F
B0
&
X
F 1 b F
Reference configuration
&
C ( X , t) F T F U2
&
B( X , t ) F 1 F T C 1
&
1
E( X , t)
C 1
2
F 1 b 1 F
1
F T e F
F C F
F C
1
T
1
1
1
&
x
Current configuration
&
b( x , t )
&
c ( x, t )
&
e ( x, t )
F FT
c 1
F T F 1
E F 1
1
1 c
2
1
1 b 1
2
V2
b 1
&
T
C p ( X , t ) F p F p Green deformation tensor
(9.223)
(reference configuration)
Fp
1
F p
T
(9.224)
with which we can define the plastic part of the Green-Lagrange strain tensor in the
reference configuration with:
&
E p ( X , t)
1 p
C 1
2
1 pT
F F p 1 strain tensor
(9.225)
(reference configuration)
F p F p
Vp
(9.226)
9 PLASTICITY
&
521
&
&
b p ( X , t ) is defined in the intermediate configuration, (see Figure 9.45). The Almansi strain
1
1
1 b p
T
1
1
1 F p F p tensor
(9.227)
(intermediate configuration)
&
Figure 9.44). Then, in comparison, the tensors E p ( X , t ) and e p ( X , t ) are interrelated, (see
Figure 9.42), by:
Ep
Fp
e p F p
(9.228)
T
1
1
1 F p F p
2
T
F p
e p F p
Fp
e p F p
T
2F p e p F p
T
2F p e p F p
T
F p e p F p
Fp
2e p
1F p
2F p
2F p
1
T
1 F p F p
T
1
F p
T
(9.229)
1
1 F p F p F p F p F p
T
F p F p 1
2E p
Ep
&
E e ( X , t)
&
e e ( x, t )
1 e
C 1
2
1
1
1 b e
Fe
Fe
Ue
(intermediate configuration)
(9.230)
1 eT
F F e 1 (intermediate configuration)
2
T
&
b e ( x , t ) F e F e (current configuration)
(9.231)
(9.232)
T
1
1
1 F e F e (current configuration)
(9.233)
&
Then, by considering the Almansi strain tensor in the current configuration, e( x, t ) , we can
define a new tensor E in the intermediate configuration as follows:
E
Fe
eF e
(9.234)
1 T
F F 1
2
components
o
E ij
1
Fki Fkj E ij
2
(9.235)
522
1 e p e p
Fks Fsi Fkt Ftj E ij
2
>
@ 12 >2E
1
2 E ste E st Fsip Ftjp E ij
2
1
Fsip E ste Ftjp Fsip Fsjp E ij
2
>
(9.236)
2 E ste E st which by substituting into
e
p
p
st Fsi Ftj
E st Fsip Ftjp E ij
(9.237)
Fp
E
Ee F p E p
(e _ p)
(9.238)
Ep
Fp
Ee F p
(9.239)
(9.240)
E ( e _ p ) E p
Let us see what we can obtain from the expression E e e p . Now, if we use the
relationships in (9.231) and (9.227) we obtain:
Ee e p
T
1
1 eT
1
F F e 1 1 F p F p
2
T
1
1 eT
F F e F p F p
(9.241)
Then, without altering the above outcome, we can apply the dot product of F e F e
T
e 1
eT
Fe
T
T
1
T
T
1
1
1 F e F e F e F e F e F p F p F e F e
2
T
1
1 F T F 1 F e F e e F e
2
T
(9.242)
Then, from the equations in (9.242) and (9.234) we can conclude that:
E
Fe
e F e
(9.243)
Ee e p
If we now consider the equations in (9.232) and (9.224), we can obtain the relationship
1
between C p and b e , i.e.:
be
Fe Fe
F F p
1
F F p
be
1 T
F C p
F F p
1
FT
1
T
F p FT
F C p
1
FT
(9.244)
(9.245)
9 PLASTICITY
1
1
Tr (b e ) 1 : F C p F T E ij Fik C kpp F jp
C :C p
1
523
F jk F jp C kpp
1
F : C p
1
(9.246)
1
Tr (C C p )
&
&
1
1 F T F 1
2
T
T
1
1 F p F e
1
1 Fe F p
2
1
F e F p
F
T
1
T
F p
F p
1
(9.247)
T
T
1
1
1
1 F e F p F p F e
1
T
T
1
1
e 1
1 F e F p F p
2
T
1
T
1
1
1 F e F e F e 2e p F e
2
1
1
T
T
1
F e e p F e 1 F e F e
2
e T
e 1
(9.248)
ee
Fe
T
e p F e
1
(9.249)
with which the Almansi strain tensor can also be defined as:
Fe
T
e p F e
1
ee
(9.250)
e ( p _ e) e e
Thus
e ( p _ e)
e ee
1
1
1
1 b 1 1 b e
2
2
1 e 1
b b 1
Then, from the equation in (9.249), and taking into account that F e
obtain:
e ( p _ e)
pT
1
1
F F p , we
1
e p F e
T
1
F p FT e p F F p
e ( p _ e) F p F T e p F
Fe
e ( p _ e)
T
(9.251)
(9.252)
and by comparing the above equation with that in (9.228) we can conclude that:
Ep
Fp
e ( p _ e) F p
We can now appreciate all the relationships obtained above in Figure 9.45.
(9.253)
524
Intermediate configuration
Intermediate configuration
&
T
2
b p ( X , t) F p F p
Vp
&
1
T
1
b p ( X , t) F p F p
&
1
1
e p ( X , t)
1 b p
&
C e ( X , t)
&
E e ( X , t)
E
&
T
C p ( X , t) F p F p
1 &
1
T
C p ( X , t) F p F p
&
1 p
E p ( X , t)
C 1
2
E (e _ p )
E
Fp
E e F p
eF e
Ee e p
Current Configuration
be
Fe Fe
e ( p _ e)
current
configuration
reference
configuration
Reference configuration
1
FT
Fe
T
e p F e
1
e ( p _ e) e e
Fe F p
F C p
1
1
1 b e
ee
Fe
Fp
&
X
1 e
C 1
2
Ue
intermediate configuration
E (e _ p ) E p
B0
Fe
&
X
B
Reference configuration
Fe
Fe
&
x
Current Configuration
&
C ( X , t) F T F U2
&
B( X , t ) F 1 F T C 1
&
1
E( X , t)
C 1
2
&
b( x , t ) F F T V 2
&
c( x, t ) F T F 1 b 1
&
1
e( x, t )
1 b 1
2
With the definition of the Jacobian determinant and the multiplicative decomposition of
the deformation gradient, we can obtain:
J
J eJ p
(9.254)
which thus defines the plastic Jacobian determinant J p and the elastic Jacobian
determinant J e , respectively, as:
1
Jp
1
det ( F p ) det (C p ) 2
Je
det ( F e )
>det(b )@
e
1
2
(9.255)
Then, the differential volume elements in the respective configurations, (see Figure 9.46),
are given by:
&
dV ( X , t )
&
J p dV0 ( X , t )
&
dV ( x , t )
&
J e dV ( X , t )
(9.256)
9 PLASTICITY
&
dA
J pF
dV
J p dV0
p T
525
&
dA
&
dA0
dV
B &
intermediate
configuration
Fe
Fp
&
dA0
B0
reference
configuration
&
current
da
configuration
dV0
&
X
&
da
Fe F p
dV
&
x
J eF e
dV
&
T
dA
J e dV
dV
J dV0
Let us consider now a differential area element in the reference configuration dA0 , (see
Figure 9.46), and by considering the multiplicative decomposition we obtain:
&
dA
J pF p
&
da
T
J eF e
T
&
dA0
&
dA
&
(9.257)
(9.258)
&
Remember that the transformation between the dA0 and da is given by Nansons formula,
&
da
&
JF T dA0 , (see Chapter 2), which can be validated by:
&
&
&
T
T
T
T
T
&
da J e F e dA J e F e J p F p dA0 J e J p F e F p dA0
&
&
e p
e
p T
T
J J F F
dA0 JF dA0
9.7.1.3
(9.259)
From the definition of the spatial velocity gradient, i.e. l F F 1 , we can introduce the
corresponding tensors that are brought about by the transformations F p and F e , that is:
l
&
( X , t)
1
F p F p
&
( x, t )
1
F e F e
(9.260)
Then, also based on the equation F 1 F 1 l , (see Chapter 2), we can introduce:
1
F p
F p
1
1
F e
F e
1
with which it is also possible to represent the spatial velocity gradient as:
(9.261)
526
D
Fe F p
Dt
F F 1
F p
F
1
F p F e F p F p F e
1
1
1
1
1
1
F e F p F p F e F e F p F p F e
1
1
1
p
Fp F e
F e F e F e
F
(9.262)
Fe l
Fe
1
l
(9.263)
( p _ e)
Note that, l e and l ( p _ e ) are defined in the current configuration, whereas l p is in the
intermediate configuration where the rate-of deformation and spin tensors are also
established:
1
l
2
Dp
pT
l
&
&
1
l
2
Wp
l
1
Fe
1
l Fe
Fe l
Fe
1
Fe
1
( p _ e)
Fe l
( p _ e)
Fe
Fe
1
(e _ e)
(9.264)
1
( p _ e)
F e , with
l e Fe
1
F F e F e F e
1
F e F e
(9.265)
pT
Fe
Fe
1
e 1
Fe l
(e _ e)
Fe
1
(9.266)
and
l
1
l Fe
Fe
1
F e F e F e F e F e l p F e
Fe
1
Fe
1
l
( p _ e)
Fe
1
1
(e _ e)
l
1
1
l e Fe Fe l
F e
Fe
1
( p _ e)
F e l
Fe
(9.267)
&
( p _ e ) sym
Fe
T
Fe
T
e T
1
l
2
( p _ e)
1 eT
F Fe l
2
T
1
F e F e l
2
1 e
C l p l
2
l
( p _ e )T
1 e
F l
2
1
Fe
T
Fe Fe Fe Fe l
l
pT
pT
F e F e F e
C e F
e 1
1
Fe
T
pT
Fe
T
Fe
F e F e F e
T
1
Ce l
pT
sym
Fe
1
(9.268)
1
1
T
l l
2
1 e
l l
2
eT
1 e
( p _ e)
l l
l e l
2
1 l ( p _ e ) l ( p _ e ) T
D e D ( p _ e)
( p _ e) T
(9.269)
9 PLASTICITY
527
T
T
1
T
1 e
1 e
l l e
F F e F e F e
2
2
1 e T e T e e T
e
e 1
F
F F F F F
De
(9.270)
If we consider that
C e
T
T
F e F e F e F e
(9.271)
1
1 e T e
F
C F e
2
(9.272)
We can now appreciate all the relationships obtained above in Figure 9.47.
Intermediate configuration
l
&
( X , t)
Dp
W
1
l
2
1
2
Intermediate configuration
1
F p F p
p
pT
l
l
pT
(e _ e)
Fe
(e _ e)
intermediate configuration
B0
&
X
l
1
F e
Fe
1
F e F e
( p _ e)
Fe l
&
( x, t )
De
1
2
D ( p _ e)
1
2
l
l
Fe
1
( p _ e)
eT
( p _ e)
l
( p _ e) T
D D e D ( p _ e)
p
reference
configuration
Fe
Current configuration
l e Fe
&
X
1
&
x
current
configuration
l
&
( x, t )
D
W
1
2
F F 1
l l
l l
T
1
2
Figure 9.47: The rate of change of the deformation tensors Multiplicative decomposition.
528
9.7.1.4
By using the Oldroyd rate, T T l T T l T , (see Chapter 4), we can define the rate of
change of an arbitrary second-order tensor T in the intermediate configuration as:
Tl
T Tl
pT
Oldroyd rate
(intermediate configuration)
(9.273)
Starting from said definition we can obtain the Oldroyd rate of the elastic part of the left
Cauchy-Green deformation tensor as:
b e l b e b e l
be
(9.274)
NOTE: In the literature, e.g. Marsden&Hughes (1983), we can find that b e is denoted by
the Lie derivative of b e .
F C p
where F
1
1
1
1
F C p F T F C p F T F C p F T
(9.275)
F , thus:
b e
l
l
1
1
F C p F T F C p F T
1
b e F C p F T b e l T
F C p
1
FT l T
(9.276)
b e l b e b e l
(9.277)
Then, by comparing the above equation with (9.274) we can conclude that the Oldroyd rate
of b e is given by:
F C p
be
1
(9.278)
FT
Note that if b l b b l T , it follows that the Oldroyd rate of b becomes the zero
tensor, (see Problem 4.1), i.e.:
b b l b b l
l b b l l b b l
e e l l
(9.279)
e l e e l
D e l
D e l
l
l e l e e l
e l e e l T
(9.280)
ee
1
1
1 b e
e e l e e e e l
T
1
1
1 F e F e is given by:
ee l l
ee l e e e e e l
eT
De
(9.281)
9 PLASTICITY
529
Thus, by starting from the equation e e ( p _ e ) e e , (see equation (9.250)), we can obtain:
e ( p _ e)
e e e
D De
1 e 1
b b 1 , we can obtain:
1 e 1 1
b
b ,
e ( p _ e)
1 e 1
b
9.7.1.5
(9.282)
D ( p _ e)
(9.283)
&
rate for T ( X , t ) in the intermediate configuration as follows:
'
Tl
T Tl
pT
Cotter-Rivlin rate
(intermediate configuration)
(9.284)
Then, if we consider both the Cotter-Rivlin rate of the Almansi strain tensor
'
e l
D e l
e e l we can
ep
Dp
Fp
T
E p F p
(9.285)
1
p T
(9.286)
F T D F F
p 1
p T
E F
p 1
Fe F p Fe
Fp
T
T
D F e
Fp
T
E F p
1
F F p , the
1
(9.287)
E (e _ p ) E p F p
E ( e _ p ) F p
1
1
Fp
T
E (e _ p ) F p
1
Fp
T
'
E p F p
1
(9.288)
ep
where we have applied the equation in (9.285). Then, the term E (e _ p ) can be obtained by
T
means of the expression of E (e _ p ) given in (9.239), i.e.: E (e _ p ) F p E e F p , thus:
E (e _ p )
and
T
T
T
F p E e F p F p E e F p F p E e F p
(9.289)
530
Fp
T
E ( e _ p ) F p
Remember that
1
Fp
T
F p E e F p F p
1
T
F p E e F p F p
T
T
1
F p F p E e F p F p
Fp
1
(9.290)
1
F p F p , so the above equation becomes:
Fp
T
E ( e _ p ) F p
1
pT
E e E e E e l
(9.291)
T
E ( e _ p ) F p
1
pT
E e E e E e l
'
(9.292)
Ee
D F e
Fp
T
E F p
Fe
D F e
Fp
T
E F p
'
1
'
e F e
(9.293)
Eee p
'
1
'
E e e p
'
(9.294)
Fp
p T
T
E ( e _ p ) F p
pT
l
Ee Ee l
(9.295)
E E , we obtain:
(e _ p )
E F
1
p 1
Fp
T
E p F p
1
l
pT
Ee Ee l
(9.296)
1 pT
F F p 1 we can obtain its rate of change as follows
2
T
1
T
1
p
p T
F , so, F E p F p 1 F p F p 1 F p F p . Then
2
2
T
1 pT
F F p F p
Fp
T
Fp
T
Fp
T
Fp
T
T
1
T
T
1
F p F p F p F p 2 l p E e 2 E e l
2
1
T
T
E F p 1 l p l p 2l p E e 2 E e l p
2
T
1
1
E F p l p 2 E e 1 2 E e 1 l p
2
1
T
1
E F p l p C e C e l p
E F p
1
Remember that C e
(9.297)
Fp
T
E F p
1
C e l
sym
(9.298)
Fp
T
E F p C e l
T
sym
F p F p
1
(9.299)
9 PLASTICITY
9.7.2
531
Remember from Chapter 3 that the Cauchy stress tensor ( x, t ) , the Kirchhoff stress
&
&
tensor ( x, t ) , the first Piola-Kirchhoff stress tensor P( X , t ) , the second Piola-Kirchhoff
&
&
&
stress tensor S ( X , t ) , the Biot stress tensor T ( X , t ) , and the Mandel stress tensor M( X , t )
are related to each other as indicated in Figure 9.48.
Remember that the tensors F and P are two-point tensors (pseudo-tensors), i.e. they are
not defined in any configuration.
F
reference
configuration
B0
current
configuration
J F T
&
X
Reference configuration
JF 1 F T
F 1 F T
&
x
Current configuration
&
( x, t )
&
( x, t ) J
M JF T F T
&
S( X , t )
&
T( X , t)
&
M( X , t ) C S
M C S
T
JR
F T P
F T F T
T
Fe
1
Fe
T
(9.300)
F 1 F e S F e
F T
F 1 F T
p 1
1
e 1
T
S F p
T
F p S F p
F 1 F e
F 1 F T
p 1
(9.301)
(9.302)
Fe
Fe
T
Fe
F e S F e
F e T
Fe
F e S
C e S
(9.303)
532
Intermediate configuration
&
X
B
Reference configuration
&
S( X , t )
&
P( X , t )
&
T( X , t)
&
M( X , t ) C S
intermediate
configuration
Fp
reference
configuration
F
B0
&
1
T
S( X , t) F e F e
&
M( X , t ) C e S
Current configuration
Fe
&
( x, t ) F e S F e
current
configuration
&
X
&
x
Current configuration
&
( x, t )
&
( x, t ) J
Starting from the equation in (9.302), i.e. S F p S F p , we can obtain the rate of
change of S as follows:
T
S
T
T
T
F p S F p F p S F p F p S F p
T
T
1
T
F p F p F p S S S F p F p F p
1
T
T
T
T
F p F p l p F p S S S F p l p F p F p
Fp
1
l p F p
(9.304)
F p . In addition, we have:
Fp
1
F p F p
1
F p
(9.305)
T
T
F p S Z p S S Z p F p
1
1
T
1
T
1
T
F p S F p F p S F p F p S F p
S
S
T
(9.306)
1
1
l p S F p
T
1
1
1
S F p
p T
F p T
Fp
S l p S S l
F p
(9.307)
T
Fp
1
S l
p T
F p
T
(9.308)
9 PLASTICITY
533
9.7.3
Fp
1
S F p
S S l
pT
, the above
T
(9.309)
y y ( F , T , B)
(9.310)
y y ( F , B)
(9.311)
y y ( F , F p , B)
Considering that F e
(9.312)
: : ( F e , B)
9.7.3.1
(9.313)
The Helmholtz free energy can be approached additively by two parts. One part is caused
by the effect of F e and the other part is caused by the effect of B , i.e.:
: : e ( F e ) : p (B)
(9.314)
One advantage of this decoupling is that we can treat the elastic part of the energy,
: e ( F e ) , in the same way as when we considered hyperelastic material which was discussed
in Chapter 8.
9.7.3.2
(9.315)
: : * ( F e , B* )
The free variables B* can be scalars, vectors, or second-order tensors which fulfills the
&
&
following law of transformation B * B (scalar), B* Q B (vector), and B* Q B Q T
(Eulerian second-order tensor). For the sake of simplicity, let us consider that B* is a
*
scalar B* B . Then, with respect to F e , remember that the deformation gradient does
not obey the second-order tensor transformation law, and is given by F e
Q F e . In
534
: : ( F e , B* ) : (Q F e , B) : (Q R e U e , B)
*
(9.316)
T
with which we
: : (R e R e U e , B) : ( U e , B)
T
(9.317)
That is, the energy function can also be expressed in terms of the right stretch tensor of the
intermediate configuration, (see Figure 9.50). Remember also that the right Cauchy-Green
deformation tensor C e is related to U e by the following equation C e
energy can also be expressed in terms of tensor C e :
: : (C e , B)
U e , so the
(9.318)
(9.319)
Remember that a scalar-valued isotropic tensor function can be written in terms of the
principal values, i.e.:
2
(9.320)
where Oe1 , Oe2 , Oe3 are the principal values (eigenvalues) of U e (right stretch tensor). Then,
as the tensors C e and b e have the same eigenvalues, (see Chapter 2), we can obtain:
2
(9.321)
(9.322)
w: e w:
:b
B
wB
wb e
(9.323)
9 PLASTICITY
:
535
w:
w: e
: b l be be l T
B
e
wb
wB
w: e w:
w:
: b e : l be e : be l
e
wb
wb
wb
w:
B
wB
(9.324)
wb
wb
wb
w:
w:
e
e
b b e results in a symmetric second-order
coaxial tensosr. Note also that
wb e
wb
w:
e
tensor, since
and b are symmetric and coaxial tensors (see subsection 1.5.9 in
wb e
Chapter 1).
w:
: l be
wb e
w:
e
l ik bkj
wbije
w: e
b jk l ik
wbije
w:
e
e b :l
b
w
(9.325)
symmetric
w:
: be l
wb e
w: e
bik l jk
wbije
w: e
bik l jk
wb eji
w:
e
e b :l
b
w
(9.326)
symmetric
In Chapter 1, (see Problem 1.16), it was proven that: if A and B are arbitrary secondorder tensors, the following is satisfied:
A :B
(9.327)
w:
e
e b
wb
sym
skew
w:
sym
: l,
:l
e be
b
w
{D
skew
w:
e
e b :D
wb
(9.328)
:
w: e w:
w:
w:
: b e : l be e : be l T
B
e
wB
wb
wb
wb
w: e
w:
w:
w:
: b 1 e be : l e be : l
B
wB
wb e
wb
wb
w: e e 1 e w:
w:
: b b b 2 e be : l
B
wB
wb e
wb
(9.329)
w:
w:
w:
e e
e 1
e
B
e b : b b 2 e b : l
wB
w
b
wb
w:
w:
w:
e e
e 1
e
B
e b : b b 2 e b : D
wB
wb
wb
Then, the rate of change of the isotropic Helmholtz free energy becomes:
w: (b e , B)
w: (b e , B) e 1 e e 1
: (b e , B) 2
b : b b D
B
e
wB
wb
(9.330)
536
w: (b e , B)
w: (b e , B) e 1 e e 1
b : b b l
B
: (b e , B) 2
e
wB
wb
C e, E e, Ue
(9.331)
Ue
Fe
&
X
: (C e , B)
intermediate
configuration
&
X
: ( E e , B)
: : (b e , B)
: ( U e , B)
Re
Fp
Re Ue
Fe
Re
: ( F , B)
e
B0
&
X
reference
configuration
Fe F p
y y ( F , B)
&
x
current
configuration
be
: : (b e , B)
9.7.4
G G ( F e , B)
(9.332)
As we can see, it is a function that depends on the same parameters as the free energy.
The yield surface is defined analogously as follows:
F F ( F e , B)
(9.333)
The plastic potential and the yield surface have to fulfill the principle of objectivity, so the
following holds:
G G ( E e , B)
F F ( E e , B)
(9.334)
We can also express these functions in the stress space. Then, if we consider that, in the
intermediate configuration, S S ( E e ) , we then have:
G G (S ( E e ), B )
F F (S ( E e ), B )
(9.335)
9 PLASTICITY
9.7.5
537
The internal dissipation Dint in the reference configuration, (see Chapter 5), is given by:
>
Dint S : E S 0 IT Z t 0
(9.336)
where Z is the Helmholtz free energy per unit mass, and : S 0 Z is the energy per unit
volume. In the isothermal process, T 0 , this dissipation becomes:
Dint S : E : : D : t 0
(9.337)
Then, if we consider the equation in (9.301), the above inequality becomes:
T
Dint : D : F e S F e : D : t 0
S : F
eT
(9.338)
D F e : t 0
: D : t 0
Dint
(9.339)
'
S : E : t 0
w:
1
w:
e 1 e
:
b e D B t 0
b
b
e
2
wb
wB
1 e e 1 w:
w:
w:
e
e
t0
B
2 e b : D 2 e b : b b
wb
2
wb
wB
: D 2
(9.340)
As the above inequality must be satisfied for any thermodynamic process, we can deduce
that:
w: (b e , B) e
b Constitutive equation for stress
wb e
In addition, by adopting C
w:
wB
(9.341)
Dint
1
1
: b e b e CB t 0
(9.342)
0 , and revert to
Note that in a process that is purely elastic (hyperelasticity), we have B
e
the following scenario b b , with which we find that the energy dissipation is equal to
obtained as that for an isotropic hyperelastic material, see subsection 8.3.1 in Chapter 8).
538
9.7.6
In order to fully describe the constitutive model we have to establish how the internal
variables evolve.
Firstly, we will define the elastic domain E in the stress space, and the yield criterion F
in terms of the Kirchhoff stress tensor (current configuration):
E ^( , C ) R :
F ( , C ) d 0`
(9.343)
where C is the scalar-valued tensor function, denoted by the isotropic hardening function
and F is assumed to be a convex function.
Next, we apply the maximum dissipation principle, which states that dissipation in the
material reaches a maximum during a change characterized by a dissipative process. Let us
consider the current state ( , C ) E which represents the current distribution of the
Kirchhoff stress tensor and thermodynamic forces in a body subjected to plastic strain. The
maximum dissipation principle requires that for a change of state, show here by
( * , C * ) E , the following must be satisfied:
> @: 12 b b
>
e 1
C C * B
t0
(9.344)
The inequality in (9.344) describes an optimization problem with constraint. Here, we can
maximize the dissipation by minimizing the negative dissipation under the constraint
'( , C ) d 0 . To this end, we define the Lagrangian:
L Dint H '
1
1
H '
: b e b e CB
2
(9.345)
1 e e 1
w'
b b H
0
2
w
w'
H
0
B
wC
0
wL
wC
be
2H
H
B
w'
wC
w' e
b
w
(9.346)
(9.347)
be
2H
w' ( , C ) e
b
w
H
B
w' ( , C )
wC
(9.348)
F ( , C ) d 0
H F ( , C ) 0
(9.349)
(9.350)
1
F C p F T , into (9.346) we can
9 PLASTICITY
F C p
1
FT
2H
w' e
b
w
F C p
1
w' e
T
2H F 1
b F
w
1
C p
539
(9.351)
w'
w' e
p 1
T
T
T
2H F 1
2H F 1
F C F F
b F
w
w
w'
p 1
2H F 1
F C
w
(9.352)
Summary
Helmholtz Free Energy (per unit
volume):
Measurement of elastic deformation:
Stress:
Isotropic hardening force:
Yield surface:
: : (b e , B)
(9.353)
Fe Fe
w:
2 e be
wb
w:
C
wB
F F (, C )
F C p
be
1
FT
(9.354)
(9.355)
(9.356)
(9.357)
Evolution equations:
w'
p 1
2H F 1
F C
w
W
w'
H
B
wC
Kuhn-Tucker condition
(loading/unloading):
H t 0
1
C p
Dissipation:
9.7.7
F ( , C ) d 0
(9.358)
H F ( , C ) 0
1
1
t0
: b b e CB
Dint
(9.359)
(9.360)
w: e
wE e
w: e
wE
1
(9.361)
Ce
w 2: e
wE e wE e
w 2: e
wE wE
(9.362)
The tensors C e and C have minor and major symmetries and they are related to each
other by:
C ijkl
1
1
1
e
Fimp F jnp C mnpq
Fkpp Flqp
1
(9.363)
540
C e : E e
(9.364)
Then, by substituting the expression of S given by (9.306), and E e given by (9.299), into
the equation in (9.364), we obtain:
T
T
C e : F p E F p C e l
T
T
F p S Z p S S Z p F p
sym
F p F p
1
(9.365)
p 1
T
Fp
1
T
T
C : E F p C e l
S
9.7.7.1
C e : F p E F p C e l
p
sym
Z
S S Z
pT
sym
F p F p
1
p T
(9.366)
Ce : E e
C e : E e p
with
Ce
w 2: e
wE e wE e
(9.367)
H
wG (S , B)
H SG
wS
(9.368)
where H is the plastic multiplier. Then, by combining (9.368) with (9.367) we obtain:
S
C e : E e p
wG (S , B)
C e : E H
wS
(9.369)
Next, to obtain the parameter H we use the consistency condition, i.e. any change in the
intermediate configuration must allow the stress state to remain on the yield surface, thus
for H ! 0 F 0 we have:
F (S , B)
wF
wS
:S
wF
wB
i 1
B i
wF
wS
wF
wS
:S
: S H
wF
wB
i 1
i 1
H H i
wF
Hi
wB i
(9.370)
0
For the sake of simplicity we have considered that B i are scalars. Then, by combining
(9.369) with (9.370) we obtain:
wF
wS
: S H
wF
wB
i 1
Hi
wF e
: C
wS
n
wG
wF
H
Hi
: E H
w
wS
i 1 Bi
n
wG
wF
wF
: C : E H
: Ce :
H
Hi
wS
wS
wS
i 1 wB i
wF
(9.371)
9 PLASTICITY
H
541
wF
: C e : E
wS
n
wF
w
G
wF
e
H i
wS : C : wS
i 1 wB i
(9.372)
Next, by substituting the above expression of H into the equation in (9.369), we find:
S
C e : E H C e :
wG
wS
wF
: C e : E
wG
wS
e
C :E
Ce :
n
wF
wS
wF
e wG
H i
wS : C : w S
w
B
i
i 1
(9.373)
Denoting by:
1
wF
w
G
e
wS : C : wS
wF
wB
i 1
H i
(9.374)
wG
wF
Ce : E K
: C e : E C e :
wS
wS
Sij
wF e e wG
C e ijkl Ekl K
C E C ijkl
wS pq pqst st
w Skl
(9.375)
wG
wF e
C e ijkl EabE ak E bl K
C pqst EabE as E bt C e ijkl
wS pq
wSkl
e
C ijkl E E K wF C e E E C e ijkl wG
ak bl
pqst as bt
wS pq
wSkl
e
C ijab K wF C e C e ijkl wG
pqab
wS pq
wSkl
E
ab
(9.376)
wG
C e ijab K C e ijkl
w
Skl
E
ab
wF e
wS C pqab Eab
pq
Thus:
S
e wG wF
C :
: Ce
w
w
S
S
e
:E
C
n
wG
w
F
F
w
e
:
C
:
H
i
wS
wS i 1 wB i
S
C tan _ ep : E
(9.377)
C tan _ ep
e wG wF
C :
: Ce
wS wS
Ce
n
wF
G
w
F
w
e
H i
wS : C : wS
B
w
i
i 1
If we consider that m
wG
wS
and n
wF
wS
(9.378)
542
Ce K Ce : m n : Ce
C tan _ ep
(9.379)
Note that m is the gradient of G in the stress space, n is the gradient of the yield surface
in the same space, i.e. n is normal to the yield surface. Then, in the case of associated flow,
F G , we obtain m n , and the elastoplastic tangent stiffness tensor is symmetric (major
and minor symmetry):
C tan _ ep
9.7.8
Ce
wF
wB H
: n n : Ce
n : Ce : n
i 1
(9.380)
In this subsection we will formulate a model for finite strain plasticity considering the J 2
flow theory with the isotropic hardening law, (see Simo&Hughes (1998)).
9.7.8.1
In this model, Simo&Hughes (1998), the Helmholtz free energy (per unit volume) is given
by:
>
~
N 1 2
N
J 1 ln( J ) Tr (b e ) 3
2 2
2
(9.381)
where J det (F ) , and b e J 2 / 3 b e is the isochoric part of the elastic part of the left
Cauchy-Green deformation tensor, N is the bulk modulus, and N G is the shear
modulus.
Then, starting from the intermediate configuration we apply a multiplicative decomposition
by means of a volumetric transformation brought about by F e
vol
vol
~
be
Fe
~
Fe
vol
F e
F~ e
vol
J e31 J e31
1
1
J e 3 F e J e 3 F e
(9.382)
Je3
Je
2
3 Fe
Fe
2
J e 3 be
(9.383)
9 PLASTICITY
F vol
volumetric plastic
intermediate
configuration
~
Jp
p vol
F
Jp
vol
~
Fp
Jp3Fp
Fe
vol
intermediate
configuration
p vol
&
X
vol
Je
Je 3 Fe
~
Je
~
Fe
current
configuration
Fe
vol
~
Fe Fe F p
vol
~
Fe Fe F p
1
~
Fe
Fe
reference
configuration
Fe
&
X
&
X
J e31
Je
1
p3
1
volumetric elastic
intermediate
configuration
B0
vol p _ e
1
~
Fp
1
~
J e3F p
J e3 J p 3 F p
F
&
X
1
p_e
543
J J
&
x
: : (b e , B )
w:
be
wb e
1 w
N
2 wb e
~
>
~
w N 1 2
N
J 1 ln( J ) Tr (b e ) 3 b e
e
2
2
2
wb
2
w
w
J 2 1 e >ln( J )@ N e J e 3 Tr (b e ) b e
wb
wb
w: (b e , B) e
b
wb e
2
2
(9.384)
544
w( J e )
wb e
w( J e J p )
wb e
1 e e 1 wJ
J b ,
2
wb e
Jp
w( J e )
wb e
Jp
>
w>ln( J )@ 1 w ( J ) 1 1 e 1 1 e 1 w Tr (b e )
J b
b ,
J wb e
J 2
2
wb e
wb e
1
w
wJ
1 1
J 2 1 2 J e 2 JJ b e
J 2be ,
2
wb e
wb
w
wb e
1 e e 1
J b
2
1 e 1
b ,
2
1,
e 2
e
J 3 Tr (b )
>
2
2 e 35 wJ
w Tr (b e )
J
Tr (b e ) J e 3
e
3
wb
wb e
2
2 e 35 1 e 1
J J b Tr (b e ) J e 3 1
3
2
Then, by incorporating the above derivatives into the equation in (9.384) we obtain:
1 w
w 2
w
J 2 1 e >ln( J )@ N e J e 3 Tr (b e ) b e
N
e
wb
wb
2 wb
2
1 2 e 1 1 e 1
2 e 5 1 e 1
J 3 J b Tr (b e ) J e 3 1 b e
N J b b N
2
2
2
3
2
N 2
1 e 2 e 1
e
e 1
e
e
J 1 b N J 3 b Tr (b ) J 3 1 b
2
3
2
N 2
1
e
~ e e 1
e 1
e
J 1 b N Tr (b )b J 3 1 b
2
3
>
>
(9.385)
>
>
2
1
1
1
N 2
~
J 1 b e b e N Tr (b e )b e b e J e 3 1 b e
3
2
N 2
~e
~e
1
J 1 1 N Tr (b )1 b
2
where
~
~
1
Tr (b e )1 b e
3
>b~ @
e dev
9.7.8.3
>
> @
N 2
~
J 1 1 N b e
2
dev
(9.386)
Fp
as an
det ( F ) det ( F e )
Je
(9.387)
>
~
N 1 e2
N
J 1 ln( J e ) Tr (b e ) 3
2 2
2
N e2
~ dev
J 1 1 N b e
> @
(9.388)
9 PLASTICITY
I b~
Ib
I C~
II b~
III b
II C~
II b
III b2
III b~
III C~
(9.389)
2
545
2
~
det (b e ) det ( J e 3 b e )
2
(9.390)
J e det (b e )
2
~
det (b e ) det ( J e 3 b e )
2
Je Je
(9.391)
9.7.8.4
I be
I C~ e
III b e
II b~ e
II C~ e
II b e
3
III b2e
III b~ e
III C~ e
(9.392)
The rate of change of the Helmholtz free energy, (see equation (9.388)), is given by
:
>
D N 1 e2
N ~
J 1 ln( J e ) b e : 1 3
Dt 2 2
2
Remember that J
J 1
C :C
2
J C 1 : E
J Tr (D)
N 1 e e 1 e N ~ e
b :1
2J J e J
2 2
J
2
(9.393)
also that:
2
D e 32 e
2 e 5 e e
e
e
J b J 3 J b J 3 b
Dt
3
~
be
Note that b l b b l
following relationship b e
2
2 e 35 e e
J J b J e 3 l
be be l e
(9.394)
:
1
N 1 e e
N ~
2 J J e J e b e : 1
2 2
J
2
2
1 e N 2 e 35 e e
N e
e 3
J
J
J
J
b
J
l
2
2 3
J e
be be l
eT
: 1
2
N 2 e 35 e
1 e
N e
e
e
e
e 3
J
J
(
D
)
J
J
(
D
)
(
b
)
J
Tr
Tr
Tr
l
2
2 3
J e
be be l
eT
: 1
(9.395)
Furthermore, if l e b e b e l
Then:
eT
: 1
e e
ik bkj
bike l jke E ij
2 l jke bkje
546
:
2
N 2 2
N e2
J 1 Tr (De ) J e 3 Tr (D e ) Tr (b e ) J e 3 2b e : D e
2
2 3
N e2
~e
~ e e
1
J 1 1 N Tr (b )1 b : D
3
> @
~e
N e2
2 J 1 1 N b
dev
:D
(9.396)
: De
9.7.8.5
Let us consider the Mises-Hubers yield condition formulated in terms of the Kirchhoff
stress tensor as:
F ( , C )
2
>V Y KB @ d 0
3
dev
(9.397)
where V Y is the yield stress, K ! 0 is the isotropic hardening modulus, and B is the
hardening parameter.
The plastic flow rule in the current configuration is given by:
e
b
dev
dev
2
H Tr (b e ) dev
3
(9.398)
2
H
3
(9.399)
F ( , C ) d 0
H F ( , C ) 0
(9.400)
(9.401)
10 Thermoelasticity
10
Thermoelasticity
S I ( x , t )
S 0 I
1
1
1 &
: D S u 2 q x& T t 0
T
T
T
1
1
1 &
S : E S 0 u 2 q 0 X& T t 0
T
T
T
Clausius-Duhem inequality
(current configuration)
Clausius-Duhem inequality
(reference configuration)
(10.1)
(10.2)
1
1
1 &
S 0 I P : F S 0 u 2 q 0 X& T t 0
T
T
T
&
&
&
Note that q x T d 0 , since the sense of the heat flux vector ( q ) is always opposite to that
of the temperature gradient ( x& T ). Thus, we can formulate the heat conduction inequality,
&
q x& T t 0
(current configuration)
&
q 0 X& T t 0
(reference configuration)
(10.3)
Then, by imposing the restriction (10.3) into the Clausius-Duhem inequality (10.1) and
(10.2) we are lead to the Clausius-Planck inequality:
E.W.V. Chaves, Notes on Continuum Mechanics, Lecture Notes on Numerical
Methods in Engineering and Sciences 4, DOI 10.1007/978-94-007-5986-2_12,
International Center for Numerical Methods in Engineering (CIMNE), 2013
547
548
1
T
&
1
T
&
Clausius-Planck Dint
inequality
1
1
S : E S 0 u t 0
T
T
(10.4)
or
(reference configuration)
&
1
T
&
1
T
Dint S 0 I ( X , t ) P : F S 0 u ( X , t ) t 0
where Dint is the internal energy dissipation (or the local entropy production), which must
be positive throughout the continuum at any point and time, i.e. D int t 0 .
Then, in a reversible process we have Dint
1
T
1
T
Dint S 0 I S : E S 0 u 0
u
S0
S : E TI
(10.5)
wu wu
:E
I
wE
wI
(10.6)
Then, by comparing the equations (10.5) and (10.6) we can draw the conclusion that:
S
wu ( E , I )
wE I 0
S0
wu ( E , I )
wI E 0
(10.7)
J
kg
(10.8)
We now need to verify that we are working with specific energy, i.e. energy per unit mass:
J
kgK
J
and then the rate of change of (10.8) is given by:
kg
Z u TI TI
(10.9)
Then, by substituting the rate of change of the specific internal energy given in (10.5) into
the above equation we obtain:
Z u TI TI
S0
S0
S : E TI
(10.10)
10 THERMOELASTICITY
549
(10.11)
wT
wE
Then, by comparing the equations (10.10) and (10.11) we can conclude that:
wZ ( E , T )
S( E , T ) S 0
wE
T 0
wZ ( E , T )
wT
E 0
I ( E , T )
(10.12)
wE
T 0
wI ( E , T )
: E
T
wT
E 0
(10.13)
E
w
wT E 0
T 0
(10.14)
S0
S : dE T dI
(10.15)
S0
wI ( E , T )
wI ( E , T )
dT
S : dE T
: dE
w
w
E
T
T 0
E 0
(10.16)
1
wI ( E , T )
wI ( E , T )
dT
: dE T
S T
wE T 0
wT
E 0
S0
The necessary and sufficient condition for du to be a total differential is guaranteed by:
w
wT
wI ( E , T )
S T
wE T 0
S0
w
wE
wI ( E , T )
E 0
wT
(10.17)
wI ( E , T )
S 0
wE T 0
(10.18)
S 0 wT
wI ( E , T )
wE T 0
wS
wT E 0
wS
wT E 0
wI ( E , T )
S 0
Pa
K
(10.19)
550
G(S , T ) Z ( E , T )
S0
S:E
J
kg
(10.20)
1
1
Z ( E , T )
S:E
S : E
S0
S0
(10.21)
Then, by substituting the rate of change of the Helmholtz free energy given in (10.10) into
the above equation, we can obtain:
wG wG
T
:S
wS
wT
S0
wG wG
T
:S
wS
wT
1
1
S : E TI
S:E
S : E
S0
S0
(10.22)
1
TI
S:E
S0
wG(S, T )
E (S , T ) S 0
wS T 0
wG(S , T )
wT
S 0
I I (S, T )
(10.23)
and if we consider the relationship between the specific Helmholtz free energy and specific
internal energy, Z ( E , T ) u ( E , I ) TI , we can obtain the following equations:
G(S, T ) Z ( E , T )
S0
u ( E , I ) TI
S:E
S : E u ( E , I ) G(S, T )
S0
S0
S : E TI
(10.24)
S0
G(S , T ) TI Specific
S:E
Enthalpy
J
kg
(10.25)
Then, by substituting u
S0
(10.26)
wH wH
:S
I
wS
wI
S0
1
1
S : E TI
S:E
S : E
S0
S0
wH wH
1
:S
S:E
I TI
wS
wI
S0
(10.27)
wH(S, I )
S 0
wS I 0
wH(S , I )
wI S 0
(10.28)
10 THERMOELASTICITY
551
Now, let us suppose that we make a change in the system characterized by the following
process T o T dT , S o S dS , I o I dI , with which the equation in (10.27) can be
expressed as follows:
dH TdI
S0
E : dS
(10.29)
S
w
wT S 0
T 0
dI
(10.30)
S0
wS T 0
wT S 0
wI
1
wI
E : dS T
: dT
T
w
S
S
T 0
wT S 0
0
(10.31)
Then, the necessary and sufficient condition for dH to be a total differential is guaranteed
by:
w
wT
wI
1
E
T
w
S
S
0
T 0
w wI
wS wT S 0
(10.32)
S
S
w
T 0
0 wT
wI
wS T 0
1 wE
S 0 wT S
(10.33)
Next, we can define a new tensor: the thermal expansion tensor, by:
wE
A
wT S 0
wI
wS T 0
S0
1
K
(10.34)
Now, we can make a summary of all the thermodynamic potentials in Table 10.1 by means
of which we can easily show that (u G) (Z H) 0 .
Table 10.1: Thermodynamic potentials.
Specific Helmholtz free
Specific internal energy
energy
u( E , I)
Z( E , T )
wu
wZ
S( E , I ) S 0
S( E , T ) S 0
T (E, I)
u G TI
wu
wI
1
S0
wE
S:E
wE
wZ
I( E , T )
wT
Z u TI
Specific enthalpy
G(S , T )
H(S, I )
E (S , T ) S 0
I (S, T )
G Z
wG
wT
S0
H TI
wG
wS
S:E
E (S , I ) S 0
wH
wS
wH
wI
1
H u
S:E
T (S, I )
S0
G TI
552
u ( E , I )
wu wu
:E
I
wE
wI
wu ( E , I )
S ise ( E , I ) S 0
wE I 0
T ( E , I ) wu ( E , I )
wI
E 0
(10.35)
and from the rate of change of the specific Helmholtz free energy we obtain:
wZ wZ
:E
Z ( E , T )
T
wE
wZ ( E , T )
S isoT ( E , T ) S 0
T 0
wE
Z
(
E
,
)
T
w
I ( E , T )
E 0
wT
wT
(10.36)
wE I 0
wZ ( E , T )
S isoT ( E , T ) S 0
wE
T 0
(10.37)
wE T 0
wT E 0
w 2Z
w 2Z
S 0
: E S 0
T
wE wE T 0
wTwE E 0
(10.38)
e
C isoT
: E MT
wS ( E , T )
wE T 0
w 2 Z( E , T )
wE wE T 0
S 0
>Pa@
(10.39)
wS T 0
wT S 0
2
w 2G
e
S w G
S 0
:
S
0
wTwS T D isoT : S AT
wS wS T 0
S 0
(10.40)
10 THERMOELASTICITY
e
where D isoT
e
C isoT
1
e
and A D isoT
1
553
: M holds.
Now, if we calculate the rate of change of S ( E , I ) , (see Table 10.1), we can obtain:
wS
wS
: E
I
S ( E , I )
wE I 0
wI E 0
w 2u
w 2u
S 0
: E S 0
I
wE wE I 0
wIwE E 0
(10.41)
C eise : E MI
wS ( E , I )
wE I 0
w 2u( E, I)
wE wE I 0
S 0
>Pa@
(10.42)
Remember that in isotropic linear elastic materials the elasticity tensor is expressed in terms
of the Lam constants as follows: C e M1 1 2NI . Likewise, we can define the adiabatic
and isothermal elasticity tensors for isotropic linear elastic materials as:
e
C ise
e
C isoT
M ise 1 1 2N ise I
(10.43)
M isoT 1 1 2N isoT I
(10.44)
where ( M adi , N adi ), ( M isoT , N isoT ) are the Lam constants for isentropic and isothermal
processes, respectively.
Specific heat at a constant strain: it is a scalar that corresponds to the heat supplied to a
unit of mass so as to achieve a unit temperature change whilst maintaining the strain
constant, ( E 0 ):
554
wI
T
wT E 0
cE
wu
wT E 0
w 2Z
Specific heat at a
T
w
w
T
T
E 0 constant volume
wI
Next, we can also check the SI unit: >c E @ T
wT
(10.45)
kgK
J
.
kgK
Specific heat at a constant stress: it is a scalar that corresponds to the heat required for a
unit temperature change while the stress is maintained constant ( P 0 ):
cS
wI
T
wT S 0
kgK
wH
Specific heat at a constant
w
T
S 0 stress
(10.46)
(10.47)
In Figure 10.1 we can appreciate the graph entropy vs. temperature for water, where we can
verify that during the phase changes there is a jump in entropy without there being any
temperature variation. In said graph we can also verify that c E c E (T ) is temperature
dependent.
We can define the latent heat tensor of change of strain as the heat that must be provided at the
material point so as to achieve a unit strain change while the temperature is maintained
constant:
LE
wI
T
wE T 0
J
K
kgK
J
(10.48)
kg
Note that in Figure 10.1 in Tsl or Tlg there is an entropy jump, i.e., we are providing heat
with no temperature change, since at these points there is a phase change.
We can define the latent heat tensor of change of stress as the heat that must be provided
at the material point so as to achieve a unit stress change while the temperature is
maintained constant:
LS
J m2
K
kgK N
wI
T
wS T 0
m3
(10.49)
kg
Note that L E and LS are symmetric second-order tensors, since E and S are symmetric
tensors too. Then, if we consider the following relationships, (see equations (10.19) and
(10.34)) we have:
wI
wE T 0
S0
wS
wT E 0
M
wI
wS T 0
S0
wE
wT S 0
(10.50)
wI
T
wE T 0
LS
wI
T
wS T 0
T wS
S 0 wT E
0
T wE
S 0 wT S 0
S0
T
S0
M
A
(10.52)
10 THERMOELASTICITY
555
I (T )
gas
wI
w5
c E (T )
T
curve tangent
liquid
Third Law of
Thermodynamics
I (T 0) I 0
solid
Tsl
Tlg
T (K )
wI ( E , T ) wI ( E , T ) wE
wI (S , T )
wI ( E , T ) wI ( E , T ) wE
:
:
T
T
T
wE
wT
wT
wT
wE
wT
wT
cS
wI
wE T 0
wE
wT S 0
LE
, thus:
T
S 0 LS ,
Note that T
S0
cS c E
(10.53)
cE
(10.54)
LS : LE
wI ( E , T ) wE
:
wE
wS
1
wI ( E , T )
: C eisoT
wE
e
C isoT
1
wI ( E , T )
wE
(10.55)
or:
LS
T
LS
1
LE
e
: C isoT
T
LE : C eisoT
1
LS
e
L E : C isoT
LS : L E
1
L E : C eisoT
1
: LE
e
LS : C isoT
(10.56)
e
LS : C isoT
: LS
(10.57)
LE
Then, by using the definition in (10.54) we can draw the conclusion that:
cS c E
S0
1
S0
e
e
LS : C isoT
: L E
: LS
L E : C isoT
T
1
e
e
and as C isoT
is a positive definite tensor, i.e. LE : C isoT
following must be met:
cS ! c E
1
(10.58)
(10.59)
Furthermore, if we return to the equation in (10.13) we can also express this as follows:
556
wI ( E , T )
I ( E , T )
wE
T 0
wI ( E , T )
: E
T
E 0
wT
S0
M : E
cE
T
T
(10.60)
T
S 0cE
M : E
(10.61)
Then, if we take the rate of change of S ( E , T ) , (see equation (10.12)), we are given:
wS
wS
: E
T
S ( E , T )
wE T 0
wT E 0
e
C isoT
: E M T
(10.62)
e
C isoT
M : E
: E M
S 0cE
e
S ( E , T ) C isoT
: E M T
T
C isoT
M M : E
c
S
0 E
(10.63)
and by comparing this with the equation in (10.41) we can conclude that:
e
C ise
C eisoT
S 0cE
MM
(10.64)
I( E , T )
&
q 0
wZ ( E , T )
The constitutive equations for simple
wE
wZ
wT
&
q 0 ( E , T , X& T )
thermoelastic materials
(10.65)
(Reference configuration)
where Z is the specific Helmholtz free energy (per unit mass), S is the second Piola&
Kirchhoff stress tensor, I is the specific entropy (per unit mass), q 0 is the heat flux
vector, E is the Green-Lagrange strain tensor, T denotes temperature, X& T is the
temperature gradient, and F is the deformation gradient.
To make the linearization of the constitutive equations, (Nowacki(1967), ilhav(1997),
Pabst(2005)), we will use the Taylor series expansion, where the following condition holds:
given a function f ( x) , said function can be approximated by using the Taylor series:
10 THERMOELASTICITY
f
f ( x)
557
1 w n f (a)
( x a ) n , applied at point a (the application point). We will now apply
wx n
n!
n 0
wS ( E O , TO )
wS ( E O , TO )
(T TO )
: (E EO )
wE
wT
Higher order
(10.66)
terms
w 2 Z ( E O , TO )
w 2 Z ( E O , TO )
| SO S0
: (E EO ) S 0
(T TO )
wE wE
wEwT
S ( E O , TO ) S 0
wZ ( E O , TO )
. Note that we have used
wE
the subscript O to indicate the variable value at the application point, so as to differentiate
this from the subscript 0 which is used to identify variables in the reference configuration.
Note that the linearized constitutive equation for stress has a linear relationship with strain
and temperature, but also considers large deformation kinematics.
Now, if we consider the equation in (10.66), we can identify these material properties:
e
The isothermal elasticity tensor ( C isoT
) (reference configuration):
e
C isoT
S0
w 2 Z ( E O , TO )
wE wE
(10.67)
If
wS ( E O , TO )
wE
T 0
wS ( E O , TO )
wT
E 0
we
then
P(F , T ) S 0
w 2 Z ( E O , TO )
w 2 Z ( E O , TO )
{ S0
wT wE
wTwE
S0
consider
the
equations
F S ,
(10.68)
wZ ( E , T )
,
wE
wZ ( F , T )
, where P is the first Piola-Kirchhoff stress tensor, we can obtain:
wF
wP ( F , T )
wT
S0
w 2 Z(F , T )
wT wF
(10.69)
0 and E O
e
C isoT
: E M (T TO )
0 , the
(10.70)
M (T TO )
with
That is, the thermal stress tensor provides stress in the absence of strain ( E 0 ).
We can also define:
(10.71)
558
The thermal stress tensor is closely linked to the latent heat tensor L E (symmetric secondorder tensor) which can be expressed as follows:
LE
TO wS ( E O , TO )
S0
wT
TO
S0
M F
TO
w 2 Z ( E O , TO )
wTwE
Latent heat tensor of change of
strain
T w 2Z
FT
O
S 0 wTwF
(10.72)
Now, we can relate the latent heat tensor ( L E ) to the thermal stress tensor ( M ) by:
w 2 Z ( E O , TO )
wTwE
TO
LE
TO
S0
(10.73)
Next, we will define some material parameters which are related to deformation, i.e. those
which are associated with the inverse of the equation in (10.70).
e
: E M (T TO ) , we can
Now, based on the stress expression given in (10.70), S C isoT
obtain the inverse relationship as follows:
e
C isoT
1
:S
1
e
e
e
C isoT
: C isoT
: E C isoT
1
: M (T TO )
1
e
C isoT
:S
1
e
C isoT
(10.74)
: M (T TO )
wE
holds. Note that if the body can deform freely, the implication is that
wS
there is no stress S 0 and here the equation in (10.74) becomes:
e
where (C isoT
) 1
e
C isoT
1
: M (T TO )
A(T TO )
(10.75)
1
:M
(10.76)
e
C isoT
1
: S A(T TO )
(10.77)
NOTE: Although we have defined the tensors L E and A , in thermal stress analysis, we
need only know the tensor M . However, as regards practice in laboratory measurement, it
is more convenient, from a practical standpoint, to obtain the thermal expansion tensor
( A ) and then we can obtain the thermal stress tensor by means of the equation
M C eisoT : A .
10.3.1.2 The Linearized Heat Flux Vector
The linearization if the heat flux vector can be obtained as follows:
&
&
&
&
&
wq 0
wq 0
q 0 ( X& T ) q 0 ( E O , TO , X& T ) q 0 O
X& T |
X& T
(10.78)
w X& T
w X& T
&
&
&
&
where we have taken into account that q 0 O q 0 ( E O , TO , X& T 0) 0 . Next, we can
10 THERMOELASTICITY
559
K0
&
wq 0
w X& T
>q& 0 @
J m
m2s K
> T @
X
> T @
&
X
W
mK
(10.79)
K
, it is easy to show that:
m
W
.
mK
(10.80)
Remember from Chapter 5 that the heat conduction inequality in the reference configuration is
&
given by q 0 X& T t 0 . Then, by substituting the linearized heat flow vector given in
(10.80) into the heat conduction inequality we obtain:
&
q 0 X& T t 0
K 0 X& T X& T t 0
(10.81)
>
K ij X& T j
X& T i t 0
X& T i K ij X& T j t 0
(10.82)
(10.83)
Note that X& T K 0skew X& T K 0skew : X& T X& T 0 , since the double scalar product
between a symmetric and antisymmetric tensor is equal to zero. Then the above equation
becomes:
X& T K 0sym X& T t 0
(10.84)
That is, the antisymmetric part of the thermal conductivity tensor has no influence on
entropy evolution or on the second law of thermodynamics, (Powers (2004)).
We will now take this opportunity to introduce the thermal diffusivity tensor as follows:
D0
1
K
S 0cE 0
m2
(10.85)
560
I( E , T ) I O
(10.86)
TO
wI ( E O , TO )
wT
TO
w 2 Z ( E O , TO )
wT wT
Specific heat
(10.87)
c E (TO )
TO
S0
M:E
c Eo
'T
TO
(10.88)
c E (T )
T
slope:
IO
TO
wI
wT
I IO
T TO
I O (T TO )
wI
wT
wZ
wZ
and I (E , T )
are linear
wE
wT
functions, the approximation of the Helmholtz free energy must present quadratic terms,
so:
Z( E , T )
wZ ( E O , TO )
wZ ( E O , TO )
(T TO )
: (E EO )
wE
wT
2
2
w Z ( E O , TO )
w Z ( E O , TO )
1
(E EO ) :
: ( E E O ) (T TO )
: (E EO )
wE wE
2
wE wT
1 w 2 Z ( E O , TO )
(T TO ) 2
2 wT wT
Z ( E O , TO )
(10.89)
10 THERMOELASTICITY
561
Then, by considering both the mechanical and thermal properties defined in the previous
subsections as well as E O 0 and 'T (T TO ) , we can express the approximation of the
Helmholtz free energy as follows:
Z( E , T ) Z O
co
'T
1
e
S (0, TO ) : E I O 'T
E : C isoT
:E
M : E E 'T 2
S0
S0
2S 0
2TO
1
SO
(10.90)
1
2S 0
E : C eisoT : E
'T
S0
M:E
c Eo
'T 2
2TO
(10.91)
S( E , T )
e
C isoT
'T
S0
1
e
E : C isoT
:E
2S 0
M :E
c Eo
'T 2
2TO
: E M 'T
1
I( E , T ) I O
&
q 0 ( X& T ) K 0 X& T
S0
M:E
(10.92)
(10.93)
c Eo
TO
'T
(10.94)
(10.95)
e
where C isoT
, M , c Eo , K 0 are the thermo-mechanical material properties, which are
obtained in the laboratory.
e
C isoT
: M L (T T0 )
Duhamel-Neumann equations
(10.96)
which is the generalized Hookes law for thermoelastic material, which is also known as the
Duhamel-Neumann equation.
562
The linear thermal stress tensor ( M L ) provides stress in absence of strain, i.e.:
M L (T TO )
with
0)
(10.97)
1
1
e
C isoT
: C eisoT : C e
1
e
C isoT
1
: M L (T TO )
(10.98)
I sym
:
e
C isoT
1
e
C isoT
1
e
C isoT
: M (T TO )
1
(10.99)
: A L (T TO )
Next, the linear thermal expansion tensor ( A L ) provides strain in the absence of stress, i.e.:
A L (T TO )
with
0)
(10.100)
Note that, here, the material point (particle) can expand freely.
&
&
Then, the constitutive equations for heat flux ( q 0 | q ) can be expressed by Fouriers law of
thermal conduction, which is given by:
&
q K T
(10.101)
c Eo
1
(T TO )
M L :
S0
TO
(10.102)
(T TO ) L
co
1
e
: C isoT
:
M : E (T TO ) 2
S
2S
2TO
(10.103)
M isoT Tr ( )1 2N isoT
(10.104)
Remember that the isotropic second-order tensor is already spherical tensor and vice-versa.
Then, in isotropic materials, the thermal tensors can be represented by the following:
ML
m1
AL
B1
K K1
(10.105)
where K is the coefficient of thermal conductivity, m is the thermal stress coefficient and
B is the coefficient of thermal expansion. We can now find the relationship between B
and m by means of the definition in (10.76), thus:
10 THERMOELASTICITY
ML
563
e
C isoT
: AL
(10.106)
(10.107)
3BN isoT
E isoT
.
3(1 2Q isoT )
Then, the constitutive equations for stress, (see equation (10.96)), becomes:
M isoT Tr ( )1 2N isoT m(T TO )1
(10.108)
When 0 we have:
m(T TO )1 m'T 1
(10.109)
which represents the spherical state brought about by the temperature change (pressure) in
isotropic materials.
Then, the inverse of the constitutive equation for stress is given by (10.98) and by
considering isotropic materials said equation becomes:
e
C isoT
1
: 'T A L
e
C isoT
1
e
C isoT
1
: B'T 1
M isoT
1
1 1
I,
2N isoT (3M isoT 2N isoT )
2N isoT
(10.110)
the above
becomes:
M isoT
1
Tr ( )1
B 'T 1
2N isoT
2N isoT (3M isoT 2N isoT )
(10.111)
with
0)
(10.112)
K1 T
KT
(10.113)
Then, the constitutive equation for entropy (Biots law), in isotropic materials, becomes:
I IO
c Eo
m
'T M L :
S
TO
IO
c Eo
1
'T m1 :
S
TO
IO
c Eo
m
'T Tr ( )
S
TO
(10.114)
564
co
1 M isoT >Tr ( )@
'T
N isoT Tr ( 2 ) m
Tr ( ) E 'T 2
2S
2
2TO
S
(10.115)
1 M isoT >Tr ( )@
N isoT Tr ( 2 )
2S
2
co
Tr ( ) E 'T 2
m
S
2TO
'T
c Eo
m
'T Tr ( )
S
TO
(10.116)
(10.117)
(10.118)
&
q KT
(10.119)
Next, we will obtain the relations between the isothermal and adiabatic Lam constants. To
do so, we will use the equations in (10.43), (10.44), and (10.64):
e
C ise
C eisoT
S 0 c Eo
ML ML
m 2T
M ise 1 1 2N ise I M isoT
S 0 c Eo
T
(m1) (m1)
S 0 c Eo
(10.120)
1 1 2N isoT I
M ise M isoT
m 2T
S 0 c Eo
N ise
and
N isoT
(10.121)
In real materials c Eo ! 0 holds with which we can conclude that M ise ! M isoT .
Now, if we start from the equations in (10.121), we can define other parameters such as the
adiabatic bulk modulus:
N ise
3M isoT m T 2N isoT
S 0 c Eo
3M isoT
2N isoT
m 2T
3
3S 0 c Eo
N isoT
m 2T
3S 0 c Eo
(10.122)
The latent heat tensors, (of change of strain L E , and of change of stress LS ), (see
equations (10.51) and (10.52)), in isotropic materials, are given by:
LE
S0
ML
S0
m1
LS
S0
AL
S0
B1
mT
1 (10.123)
where we have considered the equation m B (3M isoT 2N isoT ) 3BN isoT , (see Eq.
(10.107)). Additionally, we can obtain:
10 THERMOELASTICITY
cS c E
S0
T
LS : LE
565
S0
T
mT
m 1 :
1
S
M
N
T S 0
(
3
2
)
0
isoT
isoT
m 2T
1 :1
S 0 (3M isoT 2N isoT ) ,
3
3m 2 T
S 0 (3M isoT 2N isoT )
(10.124)
N isoT
m 2T
N isoT N isoT 3S 0 c Eo
N ise
N isoT
1
m 2T
N isoT 3S 0 c Eo
(10.125)
c Eo
1
3m 2T
c Eo S 0 (3M isoT 2N isoT )
3m 2T
o
c E S 0 (3M isoT 2N isoT )
(10.126)
m 2T
N isoT 3S 0 c Eo
N ise
N isoT
(10.127)
D
( JS ) 0
Dt
&
X& P S 0 b 0
(10.128)
&
S 0V
&
&
S 0V
&
&
S 0u
&
F S S 0 b 0 S 0V S 0 u
S S T or P F T F P T
&
&
&
S 0 u ( X , t ) S : E X& q 0 S 0 r ( X , t )
&
&
&
or S 0 u ( X , t ) P : F X& q 0 S 0 r ( X , t )
X&
1
T
Dint S 0 I S : E S 0 u t 0
(10.129)
(10.130)
(10.131)
(10.132)
566
&
q*
Su
& &
Sb( x ) dV
& &
t * ( x)
Sr dV
'T
&
q*
T*
Su
& & dV
Sb( x )
Sr dV
'T
& &
t * ( x)
n
b) Mechanical problem
Input database
Load increment
it = 0 (it-iteration)
it = it+1
t = t+t
THERMAL PROBLEM
Obtain the temperature T
'T
MECHANICAL PROBLEM
Yes
'R t tolerance?
No
Yes
New load
increment?
No
End
10 THERMOELASTICITY
567
(10.133)
(10.135)
wE
wZ ( E , T )
I( E , T )
(1 equation)
wT
&
&
q 0 q 0 ( E , T , & T ) (3 equations)
X
(10.134)
(10.136)
&
&
&
&
Sb SV SV Su
&
Su : q S r
(10.137)
(10.138)
1
T
Dint SI : S u t 0
(10.139)
1 M>Tr ( )@
N Tr ( 2 )
2S
2
m
(T T0 )
c
Tr ( ) E (T T0 ) 2
2T0
MTr ( )1 2N m(T T0 )1
&
q K T
(10.140)
(10.141)
(10.142)
(10.143)
&
Su q S r
1
T
Dint SI S u 0
(10.144)
S u TSI
(10.145)
568
Z
&
q K T
cE
(T T0 )
T0
(10.146)
(10.147)
(10.148)
Then, by substituting the expression ( S u ), given by the equation in (10.145) into (10.144),
we can obtain the following:
TSI
&
q S r
(10.149)
wZ (T )
can be obtained as follows:
wT
w 2 Z (T )
T
wTwT
(10.150)
w 2 Z (T )
T
wTwT
Sc E T
&
q S r
(10.151)
w 2Z
, we have:
wTwT
&
q S r
(10.152)
T
Finally, by substituting the constitutive equations for heat flux given in (10.148) into
(10.152) we obtain Sc E T K T S r , or:
K T S r
Sc E T
(10.153)
with which we have one equation and one unknown (temperature). Remember from
Chapter 5 that the above equation is the same as that obtained when starting directly from
the principle of conservation of energy, where we also defined the variable Q S r . Now,
to fully describe this problem (which was already discussed in subsection 5.12.1.4 in
Chapter 5) we must add the boundary and initial conditions.
(10.154)
10 THERMOELASTICITY
569
&
T0 ( X ) . In the current configuration (deformed) the stress state is characterized by the
&
&
Cauchy stress tensor ( x , t ) and by a temperature distribution denoted by T ( x , t ) .
We have defined the following tensors in the reference configuration: the Green-Lagrange
&
&
strain tensor E ( X , t ) , the second Piola-Kirchhoff stress tensor ( S( X , t ) ), and the heat flux
&
&
vector q 0 ( X , t ) . Then, in the current configuration we have defined: the Almansi strain
& &
&
&
tensor ( e ( x , t ) ), the Cauchy stress tensor ( x , t ) and the heat flux vector q( x , t ) , (see
Figure 10.5).
current configuration
reference configuration
B0
&
X
&
E( X , t)
&
S( X , t )
& &
q0 ( X , t )
&
( X , t
&
T0 ( X )
&
x
&
( x , t )
&
T ( x , t )
0) 0
&
e( x, t )
&
( x, t )
& &
q( x, t )
&
J q F T
&
&
q J 1q 0 F T
(10.155)
&
&
S 0 u ( X , t ) S : E X& q 0 S 0 r ( X , t )
(10.156)
Then, by considering the continuum without any internal heat source, the energy equation
becomes:
&
&
S 0 u ( X , t ) S : E X& q 0
(10.157)
&
q0
T
&
r 1
1 &
S 0 I S 0 X& q 0 2 q 0 X& T t 0
T
T
T
!0
Note that, here, the entropy production is only caused by the heat flux. Thus
(10.158)
570
&
1
T
S 0 I t X& q 0
(10.159)
&
X& q 0
S0
(10.160)
Dint S : E S 0 I
DT DZ
t0
Dt
Dt
(10.161)
S0
S : E IT
(10.162)
(10.163)
wT
wE
Now, by comparing the equations (10.162) and (10.163), we can conclude, as expected,
that:
S
S0
wZ ( E , T )
wE
I
wZ ( E , T )
wT
(10.164)
&
q0
K 0 ( F , T ) X& T
(10.165)
(10.166)
We can now prove the above equation is valid by starting from the following equation:
& &
q da
&
&
q 0 dA
(10.167)
&
&
where dA and da are the differential area elements in the reference and current
configuration, respectively, and which are related to each other by means of the Nansons
&
&
formula da J F T dA where J F . The following is then satisfied:
& &
q da
&
&
J q F T dA
&
J q F T
&
J q F T
&
&
q 0 dA
&
&
q 0 dA
&
q0
&
q0
q i dai
q 0 k dAk
Jq i Fki1 dAk
q 0 k dAk
Jq i Fki1 dAk
q0 k
Jq i Fki1
q0 k
(10.168)
10 THERMOELASTICITY
571
&
J q F T
K 0 X& T
J K x& T F
K 0 X& T
JF
1
Jq j Fij1
q0 I
T
JK jk T , k Fij1
K 0 IQ T , Q
K x& T
(10.169)
JFij1K jk T , k
K 0 IQ T , Q
&
& &
wT ( x ( X , t ), t ) wX q ( X , t )
wX q
wx
k
&
wT ( x , t )
{ T ,k
wx k
x& T k
&
wT ( X , t ) 1
FQk
wX q
1
T , Q FQk
(10.170)
Fqk1
X& T F 1
x& T
J F 1 K x& T
K 0 IQ T , Q
JFij1K jk T , k
K 0 X& T
J F 1 K ( X& T F 1 )
1
JFij1K jk T , Q FQk
K 0 X& T
J F 1 K F T X& T
K 0 IQ T , Q
K 0 IQ T , Q
1
( JFij1K jk FQk
)T , Q
(10.171)
and:
K0
J F 1 K F T
(10.172)
where T,QI
K 0 IQ, I T,Q
S0
q0 I , I
w 2T
wX q wX i
wK 0 iq wT
S0
> K
w 2T
wX i wX q
wK 0 iq wT
0 IQ T,Q , I
T, IQ
wT
wT wX i wX q
wX i wX q
>
S0
&
X
>K
0 IQ , I T,Q
K 0 IQ T,QI
(10.173)
wK 0
: X& T X& T with which the equation in
wT
(10.173) becomes:
I T
1 wK 0
>
S 0 wT
(10.174)
It can now be shown that the previous equation in the current configuration is given by:
I T
1 wK
S wT
(10.175)
and the rates of change of the equations in (10.164) are given by:
wZ ( E , T )
wE
S0
wZ ( E , T )
wT
which yields:
S
rate
o
rate
o
I
w 2 Z( E , T )
S 0
wE wE
: E
w 2 Z( E , T )
T
wE wT
w Z( E , T ) w 2 Z ( E , T )
:E
T
wT wT
wE wT
2
(10.176)
572
TI
w 2Z w 2Z
w wZ
w 2Z
T
T
:E
T
:
E
T
T
wT wE
wT 2
wT 2
wEwT
T
w
wT
S0
(10.177)
wZ ( E , T )
, the above equation becomes:
wE
w 2Z
wZ
T
: E T
wT 2
wE
2
w
S : E T w Z2 T
S 0 wT
wT
(10.178)
We can now define a second-order tensor denoted by the latent heat tensor of change of
strain, (see equation (10.51)), and given by:
LE
T wS
S 0 wT E 0
T
w 2Z
wTwE
(10.179)
cE
(10.180)
L E : E c E T
(10.181)
In Chapter 2 we obtained the relationship between the rate of change of the GreenLagrange strain tensor ( E ) and the rate-of-deformation tensor ( D ) as follows:
E F T D F , which is expressed in indicial notation as E ij F pi D pq Fqj with which we
T w F S F T
. Also, in Chapter 2 we obtained the relationship between
S0
wT
the second Piola-Kirchhoff stress tensor ( S ) and the Cauchy stress tensor ( ) as follows:
F S F T J , where SJ S 0 holds and by taking all of the above into consideration the
F LE F T
L E : E c E T
F L
JT w
: D c E T
S 0 wT
F T : D c E T
T w F S F T
S
wT
0
T w
: D c E T
S wT
: D c
ET
(10.182)
c E0 c(T T0 )
LE
T 0
LE
T0
(10.183)
where c is a constant while c E0 and L0E are the values of c E and L E , respectively, at
T T0 and E 0 . Then, we can represent the Helmholtz free energy (per unit mass):
10 THERMOELASTICITY
573
T
T T0
1
E : C e0 : E L0E : E
c E0 cT0 T T0 T ln
T0
2S 0
T0
c
(T T0 ) 2
2
(10.184)
C 0e : E
S 0 T T0
T0
T0
L0E
(10.185)
T
1 0
L E : E c E0 cT0 ln cT T0
T0
T0
(10.186)
Then, if we consider:
C 0e : A0
S0
T0
L0E
(10.187)
C 0e
1
: S A0 T T0
(10.188)
M 0 1 1 2N 0 I
A0
B01
S0
T0
L0E
3B 0 N 0 1
(10.189)
S0
B 0 N 0 Tr ( E ) c E0 cT0 ln
T
T0
cT T0
(10.190)
(10.191)
M0
1
Tr (S ) 1 B 0 (T T0 )1
S
2N 0
3N 0
(10.192)
574
carry out the transformation caused by F e . In this way, the deformation gradient can be
represented by:
F
Fe FR
(10.193)
Inter. Conf.
B
F
B0
&
X
intermediate
configuration
reference
configuration
Fe
current
configuration
Fe FR
&
X
&
R( X , t )
Ref. Conf.
&
x
&
R0 (X)
&
&
( x, t ); ( x, t )
Ee
ER
1 RT
F F R 1
(10.194)
E R
1 T
F F 1 we can obtain the following equation:
2
(10.195)
T
1 T
1
F F 1 F R F R 1
2
2
T
1 T
F F F R F R
(10.196)
F e F R , the above
10 THERMOELASTICITY
2 E ER
FT F FR
FR
FR
FR
Fe Fe FR
T
575
F R F e F R F R F R
T
FR
FR
F e F e 1 F R
FR
FR
T
2 E e F R
(10.197)
FR
Ee F R
Ee
E E R F R
1
(10.198)
E e
E E R F R
1
T
FR
E E R F R
1
FR
T
E E R F R
1
(10.199)
1
l R F R F R
1
F R
and
F R
1
(10.200)
Now, returning to (10.199) and by taking into account the previous equations in (10.200),
we can state that:
E e
l
RT
F R E E R F R
T
1
FR
T
E E R F R
T
1
F R E E R F R l R
1
(10.201)
We can now substitute the term E E R given by the equation in (10.198) into the above:
E e
l
RT
T
FR FR Ee FR FR
T
1
FR
T
E E R F R
T
T
1
F R FR Ee FR FR l R
1
(10.202)
E e F R E E R F R E e l R
T
T
1
T
1
l R E e F R E F R F R E R F R E e l R
l
T
RT
T
E R F R
1
FR
T
T
1
1
E R F R , we can substitute
E R by means of the
1
T
1 RT
F F R F R F R F R
1
T
1
T
1 R T R T
F F R F R F R F R F R F R
F
2
T
1
1
1
1 R T R T
1
F F R F R F R F R F R F R
F
2
1
l
2
RT
(10.203)
l R
R sym
(10.204)
{ DR
FR
T
E F R
1
l
RT
Ee Ee l R
l
R sym
(10.205)
576
Ee
FR
E E R F R
T
1
1
F R F R
DR { l
R sym
1 RT
F F R 1
&
X
B0
E
S
intermediate
configuration
FR
ER
&
X
BR
Ee
1 eT
F F e 1
Se
Fe Fe
T
&
x
Fe FR
F e
Fe
current
configuration
reference
configuration
1
F F 1
1 T
F F 1
2
F F 1 F T
multiplicative decomposition F
S
F F 1 F T
Fe FR FR
1
Fe FR Fe FR
1
T
F e F e F R
1
F e F R
T
T
FR FR
1
Se F R
T
(10.206)
Fe Fe
1
F e
T
(10.207)
J eJ R
(10.208)
Therefore, we can define the thermal Jacobian determinant J R and the elastic Jacobian
determinant J e , respectively, as follows:
1
JR
det ( F R )
det (C R 1 ) 2
Je
det ( F e )
>det(b )@
R
1
2
(10.209)
10 THERMOELASTICITY
577
Then, the differential volume elements in the respective configurations, (see Figure 10.8),
are given by:
&
dV R ( X , t )
&
J R dV0 ( X , t )
&
dV ( x , t )
&
J e dV ( X , t )
(10.210)
Remember from Chapter 2 that the material time derivative of the volume element is given
D(dV )
Dt
by
Tr ( l )dV
D (dV R )
Dt
Tr ( l R )dV R
Tr (D R ) dV R
(10.211)
&
Let us now consider a differential area element in the reference configuration dA0 , (see
Figure 10.8). Next, we will define the differential area elements in different configurations:
&
dA R
J RF R
&
da
J eF e
T
&
area element in the
dA0 Differential
intermediate configuration
(10.212)
and
T
&
dA R
(10.213)
&
&
&
dA0 (Nansons formula). Hence, the following is valid:
Remember from Chapter 2 the transformation between dA0 and da is given by the
&
equation da
&
da
JF T
&
J eF e
T
dAR
e
J eF e
T
&
dA0
R T
R T
&
dA R
J F
dV R
J R dV0
&
J R F R dA0
T
JF
&
dA0
T
J eJ RF e
T
&
T
F R dA0
(10.214)
&
dA R
&
dA0
BR&
dV R
intermediate
configuration
Fe
FR
&
dA0
B0
&
X
reference
configuration
dV0
current
configuration
F
dV
Fe FR
&
da
J eF e
&
da
dV
&
x
J dV0
dV
T
&
dA R
J e dV R
578
XR 1
dX R
R dX R dR
1
R 1 (rate)
F
dR dt
dR
1 1
F R
1 (inverse)
XR
(10.215)
where the scalar X R XR (R ) is the thermal stretch coefficient. Note, an expression for X R was
proposed by Lu&Pister(1975) which is given by:
XR
exp
B (R ) dR
R
(10.216)
where B R B R (R ) is the linear thermal expansion coefficient, and where R 0 and R show
the temperature in the configurations B0 and B , respectively.
Now, if we consider the equation in (10.215) we can express E R as follows:
ER
1 RT
F F R 1
1
XR 1 XR 1 1 1 (XR2 1)1
2
2
(10.217)
dX R
X R R 1
dR
(10.218)
dX R dR
1
2X R
1
2
dR dt
d 1 2
(X R 1)1
dt 2
FR
Ee FR
X R 1 E e X R 1 X R2 E e
(10.219)
or
1
E ER
XR2
Ee
(10.220)
Next, by substituting the expression of E R given in (10.217) into the above equation we
obtain:
Ee
1
E ER
X R2
1
1 2
E (X R 1)1
2
X R2
or
2X R2 E e
2 E X R2 1 1
1 2E
X R2 1 2 E e
X R2 1 2 E e
2E 1
(10.221)
(10.222)
(10.223)
Then, the relationship between X R and the thermal expansion coefficient, B R , is given by:
BR
B R (R )
1 dXR
XR dR
(10.224)
Therefore, we can express the rate of change of the elastic strain tensor, E e , by means of:
10 THERMOELASTICITY
E e
d 1
1
R
R
2 E E 2 E E
dt XR
XR
2
579
dX
1 dX R
1
R E E R 2 E R XR R 1
3
dR
XR
X R dR
1
XR B R R E E R E X R2 B R R 1
2
X
R
1
1
R
2
E B R 2 E E XR 1 R
E B R 2XR2 E e XR2 1 R
X R2
XR2
1
X R2
>
>
>
(10.225)
1
E B R XR2 1 2 E e R
X R2
>
1
E B R 1 2 E R
X R2
(10.226)
1
F R F R
dX R 1
R 1
1
dR
XR
1 dX R
R1
X R dR
(10.227)
1
where we have used the expressions of F R and F R given in (10.215) with which we can
obtain the rate of change of the differential volume element ( dV R ) as follows:
D (dV R )
Dt
Tr ( l R )dV R
3 dX R R
RdV
X R dR
3B R R dV R
(10.228)
Z Z e ( E e , R ) Z R (R )
(10.229)
where Z e is given in terms of the strain tensor E e and temperature R , (see Figure 10.9),
and Z R can be adjusted according to the experimental results of specific heat, (see
Lubarda(2004)). Next, we will evaluate the rate of change of the specific Helmholtz free
energy:
wZ e e wZ e dZ R
:E
Z
R
R
wR
dR
wE e
(10.230)
>
wZ e e wZ e dZ R wZ e 1
wZ e dZ R
Z
R
R
R
:E
: 2 E B R 1 2 E R
R
e
e
wR
wR
dR
dR
wE
wE X R
(10.231)
Z
1 wZ e B R wZ e
wZ e dZ R
:
E
:
1
2
E
R
2
e
wR
dR
X R2 wE e
XR wE
(10.232)
580
Then, starting from the Clausius-Planck inequality, we can express the rate of change of the
energy as follows:
t 0 S Z t S : E S IR
S : E S IR S Z
0
Z d
(10.233)
S : E IR
S0
S 0 wZ e
X R2 wE e
B R wZ e
XR2 wE e
: 1 2 E
SR
BR
&
X
S0
X R3 S R
R
0
B0
S
wZ e
wE e
Z e ( E e , R)
FR
Fe
reference
configuration
SR
Se
intermediate
configuration
Z R (R )
S0
(10.234)
wZ e dZ R
wR
dR
current
configuration
&
X
Fe FR
&
x
XR S e
J F
S 0
R
J
J RSR
FR
XR 1
X R3 ! 0
(10.235)
we can obtain
S0
X R3 S R
Now, the constitutive equations for stress (10.234) can be rewritten as follows:
(10.236)
10 THERMOELASTICITY
S 0 wZ e
XR wE
581
wZ e
XR S R
w
Ee
(10.237)
Se
or:
S
XR S e
SR
Se
with
wZ e
wE e
(10.238)
Note that the equation S X R S e could have been directly obtained by means of that in
(10.206), i.e.:
S
FR FR
1
Se F R
X R 1 X R 1
T
1
S e X R 1T
XR S e
(10.239)
Let us now suppose that Z e is a quadratic function in terms of the elastic strain tensor as
follows:
SRZe
where M R M R (R ) and N R
temperature, and:
Se
>
2
1
M R Tr( E e ) N R E e : E e
2
(10.240)
C Re : E e
with
C Re
M R 1 1 2N R I
(10.241)
where C Re
>M R 1 1 2N R I@ : E e
C Re : E e
M R Tr( E )1 2N R E
e
Ee 2N R I :
Ee
MR1 1
:
Tr ( E e )
Ee
sym
(10.242)
and if we then consider the expression of E e given in (10.221) the following is valid:
Ee
1
1 2
E (X R 1)1
2
X R2
trace
o
Tr( E e )
1
XR2
3 2
Tr( E ) 2 (X R 1)
(10.243)
Then, by substituting the above into the equation in (10.242) and by considering that
S X R S e we obtain:
S
XR S e
>
XR M R Tr( E e )1 2N R E e
1
3
1
1
X R M R 2 Tr( E ) (XR2 1)1 2N R 2 E (X R2 1)1
2
2
X R
XR
(10.244)
1
>M R Tr( E )1 2N R E @ 1 (XR 1) >3M R 2N R @1
2 XR
XR
N R (R ) ),
582
2
1
>M R Tr( E )1 2N R E @ 3 (X R 1) N R 1
XR
2 XR
(10.245)
(10.246)
B R wZ e
2
X R wE
: 1 2 E
>
BR
wZ e dZ R
S : 1 2 E
S0
dR
wR
wZ e dZ R
dR
wR
BR
wZ e dZ R
S : X R2 1 2 E e
S0
dR
wR
X R2 B R
S0
2X R2 B R
S :1
S0
wZ e dZ R
S:E
dR
wR
(10.247)
where we have used the equation in (10.223): 1 2 E XR2 1 2 E e . Note also that the
following holds:
1
S :1 S e :1
XR
>
Tr M R Tr( E e )1 2N R E e
Tr(S e )
3M R Tr( E e ) 2N R Tr( E e )
>3M R 2N R @Tr( E e )
(10.248)
3N R Tr( E e ) 3N R 1 : E e
X R2 B R
S0
S :1
3XR B R N R
3
S0
Note, the above term
2XR2 B R
S0
1: Ee
S : Ee
2X R2 B R
wZ e dZ R
dR
wR
S: Ee
S0
wZ e dZ R
wR
dR
(10.249)
wZ e
can be obtained by starting from the energy equation:
wR
SRZe
1 e
S : Ee
2
S 0Z e
XR3 e
S : Ee
2
(10.250)
where we have used the equation in (10.236). Next, we will obtain the derivative of Z e
with respect of the temperature R :
S0
wZ e
wR
X 3 wS e
3 2 dXR e
XR
: Ee
S : Ee R
dR
2
2 wR
(10.251)
S0
wZ e
wR
X 3 wS e
3 3
: Ee
XR B R S e : E e R
2
2 wR
X 3 wS e
3 2
: Ee
XR B R S : E e R
2
2 wR
(10.252)
10 THERMOELASTICITY
S0
wZ e
wR
3
e
1
3 2
1
X wS
XR B R S : 2 E (X R2 1)1 R
: Ee
2
2
2 wR
X R
3
e
3
1
X wS
B R S : E (X R2 1) Tr (S) R
: Ee
2
2
2 wR
583
(10.253)
Now, by substituting (10.252) into the entropy equation given in (10.249) we obtain:
3XR3 B R N R
S0
3XR B R N R
3
S0
1: Ee
1: Ee
2X R2 B R
S0
2X R2 B R
S0
S: Ee
wZ e dZ R
dR
wR
S: Ee
dZ R
XR3 wS e
1 3 2
e
: Ee
XR B R S : E
2 wR
S 0 2
dR
3XR B R N R
X 3 wS e
2X B
dZ R
3 2
e
XR B R S R
1 R R S
:E
dR
S0
2S 0
2S 0 wR
S 0
3
e
dZ R
1 3
e
2
2
3 wS
:E
6XR B R N R 1 4XR BS 3XR B R S X R
dR
wR
2S 0
(10.254)
e
dZ R
1 3
e
2
3 wS
:E
6X R B R N R 1 X R B R S X R
dR
wR
2S 0
3
R
e
XR
dZ
wS
e
e
:E
6B R N R 1 B R S
dR
wR
2S 0
dZ R
1
wS e
e
e
:E
6B R N R 1 B R S
dR
2S R
wR
(10.255)
(10.256)
As we have discussed in subsection 9.4.1, we can reformulate the energy equation in order
to obtain the elastic part of strain e as a free variable, i.e.:
Z Z ( e , T , B i )
(10.257)
584
Now the set of internal variables B i do not include the plastic part p of the strain tensor,
since this is already included in the free variable e
satisfied:
wZ
w e
wZ wZ
w w p
(10.258)
Finally, the rate of change of the free energy given in (10.257) becomes:
Z
wZ e wZ wZ
k
:
T
B
wT
wB k
w e
(10.259)
: D SIT SZ
Dint
1&
q T t 0
T
(10.260)
Now, by substituting the rate of change of energy given in (10.259) into the above
inequality and by considering that D E | e p (in a small deformation regime), we
obtain:
1&
wZ
wZ wZ
k q T t 0
: ( e p ) SIT S e : e
T
B
wT
wB k
T
w
Dint
(10.261)
wZ e
wZ
1&
p
k q T t 0
S e : : S I
T Ak B
wT
T
w
S
where Ak
wZ
are the thermodynamic forces. As the above inequality must be valid
wB k
(10.262)
wZ ( e , T , B i )
w e
wZ ( e , T , B i )
wT
Ak
S
wZ
, I
w e
wZ
. Then, we
wT
wZ ( e , T , B i )
wB k
(10.263)
Dint
k
: p Ak B
1&
q T t 0
T
(10.264)
which we can the split into mechanical and thermal parts, i.e.:
where:
Dmechanical
Dint
k t0
: p Ak B
Dmechanical Dthermal
and
Dthermal
(10.265)
1&
q T t 0
T
(10.266)
10 THERMOELASTICITY
Dmechanical
&
k t0
^
Ak ` p B
T
Dthermal
and
b4
B4
585
q
^,
T ` t 0
B
(10.267)
b
where B4 and B include all forces, and b4 , b include all variables related to flux. Then,
&
wK
w p
Ak
wK
k
wB
wK
&
q
w
T
(10.268)
wK *
w (T )
(10.269)
&
q
T
wK *
w
k
B
wK *
wAk
;
&
&
&
q
q
k T M p , B
k ,
Sup & : p Ak B
T
T
q
k ,
p ,B
(10.270)
H
w)
wB 4
H t
(10.271)
and, according to the maximum plastic dissipation principle, the following is satisfied:
p
H
wF
w
(10.272)
where H is the plastic multiplier, F is the yield surface (which is a convex function) and
the plastic flow direction is normal to the surface F , (see Chapter 9).
11 Damage Mechanics
11
Damage Mechanics
11.1 Introduction
The term Continuum Damage Mechanics has been used to models materials which are
characterized by loss of stiffness, i.e. by a decrease in their stiffness modulus. Damage
models have also been used to simulate different materials (fragile and ductile), which are
fundamentally characterized by irreversible material degradation. Physically speaking, we
can describe the degradation of mechanical material properties as processes in which the
initiation and growth (propagation) of micro-defects such as micro pores and microcracks
take place.
In the pioneering work of Kachanov (1958) the concept of effective stress was introduced,
and by using continuum damage he solved problems related to creep in metals. Rabotnov
(1963) gave the problem physical meaning by suggesting we measure how the sectional area
has reduced by means of the damage parameter. Nowadays, Continuum Damage
Mechanics has become an important tool and is a consistent theory based on irreversible
thermodynamic processes (the Clausius-Duhem inequality). Thermodynamic formalism
was developed by Lemaitre&Chaboche (1985) and among important contributors to our
knowledge about damage mechanics we can include: Mazars (1986), Mazars&PijaudierCabot (1985), Chaboche (1979), Simo&Ju (1987 a,b), Ju(1989), Oliver et al. (1990) and Oller
et al. (1990).
The continuum damage models, from a computational point of view, are very attractive
since these present simple algorithms and are satisfactory for solving large problems.
In this chapter we will present some basic damage models used to study the failure
mechanism after which we can develop more complex ones.
587
588
material point
micro
crack
V - effective stress
V - apparent stress
s
V
sV
(11.1)
The equation (11.1) can also be rewritten without altering its outcome as follows:
V
s
V
s
sd
ss
s s
V
1 1
V 1 V
s
s
s
(11.2)
sd
represents the amount of the original section which is corrupted,
s
11 DAMAGE MECHANICS
589
sd
sd
s
0V
sd
sd
s
1 V
The amount s d depends on the stress state V or indirectly on H . The dimensionless ratio
sd
represents the damage variable and is denoted by d
s
sd
. Then, the equation in (11.2)
s
1 d V
0 d d d1
(11.3)
EH
(11.4)
where E is Youngs modulus. Then, by substituting (11.3) into (11.4) we can obtain the
constitutive equation for stress in the one-dimensional isotropic damage model:
V
1 d E H
0 d d d1
(11.5)
We can now verify that as the damage variable evolves, the state no longer returns to its
original value. Physically speaking, we can interpret this as once the material has suffered
damage this will be permanent. Hence, we can conclude that d t 0 , which characterizes an
irreversible process. Now, the equation in (11.5) can still be written as:
V
E sec_d H
E sec_d
with
1 d E
(11.6)
where E sec_d is the damage secant stiffness modulus with which we can observe that the damage
variable can be interpreted as a measure of the loss of stiffness modulus of the material.
In general, materials have a yield stress that separates the elastic (reversible process) from
the inelastic zone (irreversible process). In the strain space, we can represent the elastic
limit by the variable H 0 , (see Figure 11.2), in which the damage process has not yet begun,
i.e.:
d
if
H H0
(11.7)
s d
0;
590
1 d E H
d
(0 d d d 1)
(11.8)
H H0
if
Now, by starting from the above equation we can obtain the energy equation in the system
as follows:
1
HV
2
1 d 1 H E H
2
1 d : e
(11.9)
:e
Loading(damage)
VY
Elastic limit
1
H0
E sec_d
(1 d ) E
V
V
effective stress V
V (1 d ) V
11 DAMAGE MECHANICS
591
(11.10)
or explicitly as follows:
(11.11)
where : e ( ) is the elastic strain energy density, which is a function of strain only, and C e
is the elasticity tensor (or elastic stiffness tensor).
11.2.2.2 Internal Energy Dissipation and the Constitutive Equations
The damage model has thermodynamic consistency, and so, entropy inequality is fulfilled.
One way to express this entropy inequality is by means of the alternative form of the
Clausius-Planck inequality, (see Chapter 5), which is expressed by:
>
J
m3
Dint : D I T : t 0
(11.12)
Note that the terms : D , I T , : have the unit of energy per unit volume (density
energy). In a small deformation regime D | holds, and by considering the isothermal
process we have T 0 , so, the equation in (11.12) becomes:
Dint : : t 0
(11.13)
w: w:
: , d
:
d
(11.14)
Then, the rate of change of the free energy : : , d can be evaluated as follows:
w
wd
Next, by substituting (11.14) into the internal energy dissipation given in (11.13) we obtain:
Dint : : (, d ) :
w: w:
:
d
wd
w
w: w:
w : wd d t 0
(11.15)
Note that the above inequality must hold for any admissible thermodynamic process, so, let
w:
must also be true for any process. Additionally, if we have a process such that o , the
w:
holds with which we
only way for the entropy inequality to be satisfied is when
w
obtain the constitutive equation for stress. Thus, the entropy inequality becomes:
Dint
w: w:
w : wd d
0
w:
d t0
wd
(11.16)
592
1 d : e , we obtain w:
: e , thus
wd
Dint : e d t 0
(11.17)
d t 0
(11.18)
where by definition : e t 0 . Then, to satisfy the inequality (11.17), the rate of change of the
damage parameter must satisfy:
Then, by means of thermodynamic considerations we can draw the conclusion that:
w:
w
d t 0
(11.19)
We can also express the rate of change of the Helmholtz free energy by means of the
equation in (11.11), i.e.:
: : e 1 d d : e
1 d : C e : d : e
: d : e
1
e : : C e :
: Ce : : C
2
: e
1
2
(11.20)
: C e : , was obtained as
(11.21)
where C e 0 , since C e is constant, and as the elasticity tensor features major symmetry
e
( C eklij C ijkl
), the equation in (11.21) becomes:
: e
1
e
H ij C eijkl H kl H ij C ijkl
H kl
2
H ij C eijkl H kl
: C e :
12 H
:
e
kl C klij H ij
e
H ij C ijkl
H kl
(11.22)
1
:
1 d
C e : is fulfilled.
Then, starting from the equation in (11.11) we can obtain the stress by taking the derivative
of the strain energy with respect to strain, i.e.:
V ij
w: , d
wH ij
w
1 d 1 H pq C epqkl H kl
wH ij
2
1 d 1 C epqkl H kl
wH pq
H pq
wH kl
wH ij
1 d 1 C epqkl
2
>
w
H pq H kl
wH ij
wH ij
w 1 H H
w>1 H H @
1 d 1 C epqkl H kl 2 pq qp H pq 2 kl lk
wH ij
wH ij
2
1 d 1 C epqkl 1 H kl E pi E qj E qi E pj 1 H pq E ki E lj E li E kj
2
2
2
1
1 1
1
e
e
C ejikl H pq C epqij C epqji 1 d H kl C ijkl
1 d H kl C ijkl
H pq C epqij
2
2 2
2
>
(11.23)
where we have taken into account the minor symmetry of the elasticity tensor, i.e.
e
C ijkl
C ejikl , C epqij C epqji . Note also that the indexes p , q are dummy indexes, so we can
11 DAMAGE MECHANICS
593
exchange them for k and l without altering the expression. Additionally, by taking into
e
account the major symmetry of the elasticity tensor, C ijkl
C eklij , we obtain:
1 d C eijkl H kl
V ij
(11.24)
w: , d
w
1 d C e : 1 d
(11.25)
(11.26)
and C e is the elasticity tensor (fourth-order definite positive tensor) which contains the
elastic mechanical properties. Remember that C e can be represented in terms of the Lam
constants ( M , N ) as follows:
Ce
M1 1 2 N I
e
C ijkl
ME ij E kl NE ik E jl E il E jk
(11.27)
where 1 is the second-order unit tensor, and I { I sym is the symmetric fourth-order unit
tensor, whose components are expressed in terms of the Kronecker delta ( E ij ) as follows:
(1) ij
E ij
if
if
iz j
sym
I ijkl
{ (I) ijkl
ijkl
1
E ik E jl E il E jk
2
(11.28)
Then, by analyzing the constitutive equation in (11.25) we can put in evidence the
following sentences:
x Since the damage parameter is a scalar, the stiffness degradation is isotropic;
x We can calculate the stress immediately once we know the current values of
(strain) and d (internal variable);
x We can interpret the equation in (11.25) as the sum of elastic and inelastic parts, i.e.:
1 d C e :
e
C
: d
C
e
:
elastic
e i
inelastic
(11.29)
We can then define the elastic-damage secant stiffness tensor for the isotropic damage
model as:
C sec_d
(11.30)
Let us now consider a uniaxial case, (see Figure 11.4), where the material is loaded until the
stress state reaches the point P represented in Figure 11.4, after which unloading occurs,
with the unloading path being that indicated by the slope E sec_d (1 d ) E defined in
Figure 11.4.
11.2.2.3 Ingredients of the Damage Model
The damage constitutive model is completely determined when the damage variable d t is
known at each time step t of the loading/unloading process. Then, we can define the
following elements of the constitutive equation:
594
The damage surface and damage criterion. The damage surface defines the elastic
limit, and the damage criterion establishes when the material is in a loading or in a
elastic process, and;
Dissipated energy
VY
E
E sec_d
(1 d ) E
The norm is a measure of distance and so is a scalar. Next, we will define a simple norm in
the stress space denoted by t (equivalent stress), and in the strain space denoted by t .
The latter is also known as the equivalent strain:
1
; t
2: e
C e 1
: Ce :
Ce
: Ce :
(1 d )t
(11.31)
Note that t and t are surface equations (ellipsoids) that characterize the stress state at
the current point (see Figure 11.5). The proof of (11.31) now follows:
1
: Ce :
: Ce :
1 d :
:
1 d 2 : 1 d
:
t
1 d t (11.32)
In order to better describe material behavior, others norms will be introduced (see
subsection 11.2.4).
The Damage Criterion
Next we will define the damage criterion in the stress and strain space:
F (t , q ) t q (r ) d 0 and G(t , r ) t r d 0
stress space
strain space
(11.33)
11 DAMAGE MECHANICS
595
starts to fail (initial damage) when the energy norm exceeds the value r0 . Later we will
relate the variables r and q to the damage variable.
H2
G(t , r ) 0
V2
(1 d )t
t
F (t , q) 0
V1
H1
V3
H3
E :
E :
G(t , r ) d 0 `
(11.34)
F (t , q ) d 0 `
(11.35)
When it holds that F (t , q) 0 , in the stress space, the stress state is on the surface as
indicated in Figure 11.5(b).
The stress space ( E ), (see Eq. (11.35)), can be decomposed into the inner domain
int E (when the stress state is inside the surface), and other by the surface itself, wE .
We can define then the elastic region in strain and stress respectively as:
intE :
G(t , r ) 0 `
intE :
wE :
G (t , r ) 0 `
E int E wE
F (t , q ) d 0 `
F (t , q) 0 `
int E wE
(11.36)
(11.37)
(11.38)
Note that int E is the same as F (t , q) 0 which describes the elastic region, and wE
is the damage surface. Note that when the stress state is at a point inside of the space E it
will also be inside the space E , and when the stress state is on the surface wE it will also
be on the surface wE . Hence, we can use either the stress or strain space to describe how
the damage evolves, proof of which follows:
t q(r ) d 0
(1 d )t (1 d )r d 0
t r d 0
(11.39)
596
Said damage evolves when the norm t exceeds the maximum value reached by r . Then,
considering (11.33) and (11.31) we can also conclude that:
q (r ) (1 d )r
(11.40)
In uniaxial cases, damage starts when t exceeds the first damage threshold value r0 .
Then, from the equation in (11.31) and by means of Figure 11.2, we can obtain:
: C e : uniaxial
o t
t r0 0
H 0 EH 0
VY
E
H0 E
VY
E
(11.41)
VY
r0
2( r ( 2 ) m r0 )
VY
3( r ( 3) m t )
1 d
4( r ( 4 ) m t )
5( r ( 5) m r ( 4 ) )
H0
t
2
H1
6
5
r0
2
1
t
!0
r0
r
3
2
4, 5
H1
r0
t
r0
r3
r4
r5
11 DAMAGE MECHANICS
597
The Internal Variable Evolution Law. The Kuhn-Tucker and Consistency Conditions
The constitutive equation described above uses three types of variables, namely: the free
variable ^ ` ; the internal variable ^r`; the dependent variables ^: (, r ), (, d ), d (r )` .
Now, to establish how the internal variable r evolves, let us take the example described in
Figure 11.6. As we can observe, the discretized r between points 2-3 and 3-4 are positive
and between points 1-2 and 4-5 are equal to zero, so we can conclude that r is a
monotonically increasing function, i.e.:
r t 0
(11.42)
Graphically, we can see in Figure 11.6 how the variables r and t evolve. Furthermore,
we can also verify that in the range between the points 4-6 G(t , r ) t r 0 holds, i.e.
there is an elastic regime.
Thus, we can establish that at time t , r t is given by the following equation:
rt
max r0 , max
,t s
sf ,t @
(11.43)
w:
wd
w: w: wd (r ) wr
:
w
wd wr wt
wd (r )
w:
: : e
r
w
wr
(11.44)
wd (r )
w:
: : e
r t 0
w
wr
wd (r )
w:
e wd ( r )
r : e
r t 0
w : :
wr
wr
Dint : : (, r ) :
(11.45)
If we compare the above inequality with the one obtained in (11.17) we can conclude that:
d
wd (r )
r H(t , d ) r
wr
(11.46)
where H is the continuum hardening/softening modulus. The evolution laws for r and
for d (damage variable) are then given by:
r [ (, r )
d
[ H(t , d )
(11.47)
598
inside the damage surface, i.e. G(t , r ) t 0 , and if in the next loading step G(t , r ) t 't 0
is satisfied, this implies that G 0 r [ (, r ) 0 . We can gather these previous
conditions by means of the loading/unloading condition, also called the Kuhn-Tucker conditions:
[ t 0
G(t , r ) d 0
(11.48)
(11.49)
If we are undergoing loading, this implies that [ ! 0 , then by means of the Kuhn-Tucker
conditions G(t , r ) 0 must be fulfilled. Here, the value of [ can be obtained by means
of the consistency condition:
G(t , r ) G(t , r ) 0
t
r
(11.50)
G 0
G 0
G 0
[ 0
d
(elastic)
[ 0
d
(unloading)
[ 0
[ ! 0
d
(neutral loading)
(11.51)
(loading)
d ! 0
wG (t )
, where we have introduced the scalar function G
wt
F t F q d 0
tt0
;
;
tt0
(11.52)
[ t 0
G (t , r ) d 0
wG (t , r )
d [
wr
(11.53)
(11.54)
[ G (t , r ) 0 The consistency condition
(11.55)
d (r ) 1
q(r )
r
(11.56)
11 DAMAGE MECHANICS
599
(11.57)
0 d d d1
(11.58)
Note that with the new definition of the damage parameter given in (11.56), we can
restructure the equation in (11.46) as follows:
q(r )
r
wq (r )
q (r ) wr
r
r2
d (r ) 1
d
wd (r )
r
wr
w
wr
q(r )
1 r r
(11.59)
d
q( r ) H d (r )
r
r2
(11.60)
wq (r )
, which is the hardening/softening
wr
H d (r ) r ;
wq (r )
defines the hardening/softening parameter, thus:
wr
r >r0 (d
q >r0 , a @ ;
0), f(d 1) ;
q0
r0
VY
E
(11.61)
Perfect Damage
Damage with Softening
H d (r ) 0
H d (r ) 0
(11.62)
q(r )
r H d (r r ) 0
0
0
rd0
(11.63)
r ! r0
Then, taking into account the equation in (11.56) we can still state that:
d 1
q
r
1
r d r0
r0
r
Hd 1
r0
r
! 1
(11.64)
r ! r0
600
q (r )
Hd ! 0
r0
Hd 0
r0
r (d
1)
The exponential law is described by Figure 11.8. Then we can express q (r ) as follows:
q (r ) q f q f r0 exp
r
A 1
r0
(11.65)
A!0
with
q f r0
r0
exp
r
A 1
r0
(11.66)
q (r )
qf
q f ! r0
r0
q f r0
rf
r0
: , r >1 d (r )@ : e
Damage parameter
d (r ) 1
Evolution law
r [
Damage criterion
G , r t r
w:
w
q
r
with
:e
q >r0 , a @, a z f ;
1 d 1 d C e :
r >r0 , f
VY
r0 r t 0
E
: Ce : r
1
: Ce :
2
d >0,1@
(11.67)
(11.68)
(11.69)
(11.70)
(11.71)
11 DAMAGE MECHANICS
Hardening Law
q
H d (r ) r
G0 ;
[ G 0
Loading/unloading condition
Consistency condition
[ t 0
601
q c( r ) d 0
(11.72)
[ G 0
(11.73)
(11.74)
1 d C e :
d
(11.75)
b)
1 d C e Y C e
C tan_d
Y (1 d )
where
(11.76)
wd wr
wr wt
wd
r
wr
d (r ) becomes:
wd
t
wt
(11.77)
: Ce :
1
: Ce :
2
1
o t
: Ce :
: C
1
2
: C e :
: : C e :
1
: C e :
:
(11.78)
wd 1
:
wt t
(11.79)
Then, taking into account the equations (11.79) and (11.75), we can find the relationship
between the rates of stress and strain change:
1 d C e : d 1 d C e :
wd 1
e
:
1 d C
w
t
t
wd 1
:
wt t
(11.80)
602
wd 1
e
1 d C
w
t
t
wd 1
wr r
(11.81)
q(r ) H d r 1
r
r2
q(r ) H d r
r3
wd
wr
(11.82)
q(r ) H d r
.
r2
Then, by substituting the equation in (11.82) into that in (11.81) we can obtain C tan_d in
terms of q and r :
q(r ) H d r e
e
C
(1 d ) C e
:
:C
C tan_d
(11.83)
Now, the general equation for the elastic-damage stiffness tensor C tan_d (symmetric
fourth-order tensor) is given by:
C tan_d
where, K
q(r ) H d r
and Y
r3
1 d .
r 0oK 0
t I
: Ce
1
(1 d ) :
(11.85)
We can also define the energy norm of the strain tensor (also known as the equivalent
strain), proposed by Simo&Ju(1987), (see equation (11.31)):
t I
: Ce :
2: e
(11.86)
11 DAMAGE MECHANICS
603
To better illustrate this model, let us consider the state of plane stress ( V i 3 0 ). In this
case, the yield surface is represented by an ellipse, (see Figure 11.9), where V Y ! 0 is the
stress limit for tension and compression and the damage surface evolves symmetrically.
V2
t r0
VY
VY
E
1
VY
Elastic
region
V1
VY
VY
b) Stress-strain curve
Figure 11.9: Damage surface in 2D and the uniaxial stress-strain curve for model I.
11.2.4.2 The Tension-Only Damage Model Model II
The tension-only damage model does not take into account failure by compression, i.e. the
material can only fail by tension and here we can define the following stress field:
where x
def
x x
2
(11.87)
if x 0
if x t 0
x
Figure 11.10: Ramp function.
Now, by means of spectral representation, we can represent the stress tensor in terms of
eigenvalues (principal stresses) and eigenvectors as follows:
( a ) n ( a )
an
(11.88)
a 1
thus:
n ( a ) n ( a )
(11.89)
a 1
Note, the relationship between the real and effective stress remains valid, i.e.:
(1 d )
(11.90)
604
Then, the norm for the isotropic damage model defined previously becomes:
2: e
: Ce :
(11.91)
t II
1
1
: Ce :
2
(1 d )
1
:
: Ce :
1
1
: Ce :
(1 d )
(11.92)
t II
1
(11.93)
Finally, in Figure 11.11 we can visualize the damage surface for two-dimensional cases
(2D).
11.2.4.3 The Non-Symmetrical Damage Model Model III
The non-symmetrical damage model is useful to simulate materials, such as concrete,
whose tension domain differs with respect to compression. This model uses the following
norm:
1 R
R
t III
1
: Ce :
(11.94)
where the parameter R is the weight factor dependant on the stress state which is given
by:
3
V
i
i 1
3
(11.95)
Vi
i 1
The parameter n is defined by means of the ratio of the compression elastic limit V Yc to
the tension elastic limit V Yt , i.e.:
n
V Yc
(11.96)
V Yt
t r0
VY
E
1
Elastic region
VY
V1
b) Stress-strain curve.
Figure 11.11: Damage surface in 2D and the uniaxial stress-strain curve for model II.
11 DAMAGE MECHANICS
605
V2
t r0
V Yt
V Yt
V Yt
V Yc
V1
Elastic
region
V Yc
V Yc
a) Norm in the principal stress space-2D.
n V Yt
b) Stress-strain curve.
Figure 11.12: Damage surface in 2D and the uniaxial stress-strain curve for model III.
M1 1 2 N I
QE
QE
11
I
(1 Q)(1 2Q)
(1 Q)
N
1
1 2 N I 1 1
3
volumetric
part
(11.97)
isochoric part
Q(1 d ) E
Q(1 d ) E
11
I
(1 Q )(1 2Q )
(1 Q)
(1 d )C e
QE sec_d
QE sec_d
11
I
(1 Q)(1 2Q)
(1 Q)
Note that, in this model the damage variable affects only one of the mechanical parameters,
namely, the Youngs modulus. We can also verify that the same damage parameter equally
affects both the spherical and deviatoric part:
C sec_d
(1 d )C e
(1 d ) N1 1 (1 d )2 N I 1 1
3
(11.98)
Another model described by Carol et al. (1998) generalizes the isotropic damage model by
considering independent degradation of the spherical and deviatoric parts and because of
this the model requires two independent damage variables.
Now, the elasticity tensor components can be expressed by means of their spherical and
deviatoric parts as follows:
e
C ijkl
1
1
NE ij E kl 2 N E ik E jl E il E jk E ij E kl
3
2
(11.99)
606
1
D
E ij E kl and Pijkl
3
V
Then, with Pijkl
C eijkl
1
V
E ik E jl E il E jk Pijkl
, the above equation becomes:
2
V
D
3NPijkl
2 N Pijkl
Ce
3NP V 2 N P D
(11.100)
Let us now consider that the material parameters N and N can be degraded by means of
the variables d V and d D , respectively, and according to the following equations:
N (1 d V ) N 0
; N (1 d D )N 0
(11.101)
V
D
3(1 d V ) N 0 Pijkl
2 (1 d D )N 0 Pijkl
e _V
(1 d V )C ijkl
(1 d D )C eijkl_ D
(11.102)
e_ D
C ijkl
V
3N 0 Pijkl
N 0 E ij E kl
D
2N 0 Pijkl
(11.103)
1
2N 0 E ik E jl E il E jk E ij E kl
3
1
1
: Ce :
: 3NP V 2 N P D :
2
2
: e _ vol : e _ dev
1
1
: 3NP V : : 2 N P D :
2
2
(11.104)
1
1
: 3NP V :
: Ce _V :
2
2
e
e _ vol
: e _ dev
: ( ) :
1
1
D
e_D
: 2N P :
:C
:
2
2
(11.105)
>
1
1
: C sec_d :
: (1 d V )C e _ V (1 d D )C e _ D :
2
2
1
1
(1 d V ) : C e _ V : (1 d D ) : C e _ D :
2
2
(1 d V ): e _ vol (1 d D ): e _ dev : vol (, d V ) : dev ( , d D )
: vol
(11.106)
: dev
: ( , d V , d D ) (1 d V ) : C e _ V : (1 d D ) : C e _ D :
: vol
: dev
(11.107)
11 DAMAGE MECHANICS
607
C sec_d :
(1 d V )C e _ V : (1 d D )C e _ D :
(1 d V ) sph (1 d D ) dev
(11.108)
where sph , dev are the spherical and deviatoric effective stresses, respectively and where
the following is valid:
sph
dev
(1 d V ) sph
(1 d D ) dev
sph dev
(11.109)
(1 d V )C e _ V : (1 d D )C e _ D :
(1 d )C
e _V
sph
(1 d )C
e_D
(11.110)
dev
Then, the relationship between stress and strain in rate is given by:
sph
dev
C tan _ d :
C
tan _ d
tan _ d
:
:
sph
sph
C
dev
tan _ d
sph
dev
dev
:
C tan _ d : sph
C tan _ d : dev
(11.111)
Then, by evaluating the rate of change of the strain energy function given in (11.106),
: (1 d V ): e _ V (1 d D ): e _ D , we can obtain:
: : e _ V (1 d V ) : e _ V d V : e _ D (1 d D ) : e _ D d D
: e _ V (1 d V ) : e _ D (1 d D ) : e _ V d V : e _ D d D
(11.113)
>(1 d
Note that : e _ V
: C e _ V : and : e _ D
:
)C e _ V : (1 d D )C e _ D : :
(1 d ) : C e _ V : (1 d D ) : C e _ D :
V
(11.114)
: C e _ D : , thus:
(1 d V ) : C e _ V : (1 d D ) : C e _ D :
(1 d V ): e _ V (1 d D ): e _ D
(11.115)
Then, together the equations (11.115), (11.113), and the internal energy dissipation given in
(11.112), yields:
608
Dint : : t 0
(1 d V ): e _ V (1 d D ): e _ D : e _ V (1 d V ) : e _ D (1 d D ) : e _ V d V : e _ D d D t 0
: e _ V d V : e _ D d D t 0
(11.116)
Since (11.116) must be satisfied for any admissible thermodynamic process, it follows that:
d V t 0
d D t 0
(11.117)
t V
2: e _ V
sph :
sph : sph
sph : C e _ V : sph
(11.118)
t D
2: e _ D
dev :
dev : dev
dev : C e _ D : dev
(11.119)
t V
t D
sph
t D
:C
e _V
dev
:
sph
sph
: C e _ V : sph
t V
sph
: sph
t V
sph
:
(11.120)
(11.121)
w w V
w D
: V d
d
w
wd
wd D
w
w sph
w
: V d V
wd
w
sph
dev
w sph dev
w
w D
: ( ) V d V
d
w
wd
wd D
w D
(11.122)
: dev
d
D
wd
sph
w
wd D
wd V V
t
wt V
d D
dev
(11.123)
and
d V
wd V wr V
wr V wt
wd V V
r
wr V
wd D wr D
wr D wt
wd D D
r
wr D
Then, we can express the rates of change sph and dev as follows:
wd D D
t
wt D
(11.124)
11 DAMAGE MECHANICS
sph
w sph
w
: V d V
w
wd
609
(1 d V )C e _ V : sph
(1 d V )C e _ V : sph
wd V 1
sph :
wt V t V
wd V V
t
wt V
wd V 1
V
e _V
sph sph :
(1 d )C
V
V
wt t
(11.125)
wd V 1
V
e _V
sph sph : sph
(1 d )C
V
V
w
t
t
and
dev
w dev
w D
:
d
w
wd D
wd D D
t
(1 d D )C e _ D : dev
wt D
wd D 1
(1 d D )C e _ D : dev
dev :
wt D t D
wd D
D
e_ D
(1 d )C
wt D
wd D
D
e_ D
(1 d )C
wt D
dev dev :
D
t
1
dev dev : dev
D
t
(11.126)
wd V 1
wd D 1
D
e_D
dev
dev
sph sph :
(1 d V )C e _ V
(1 d )C
D
D
V
V
wt t
wt t
(11.127)
and by comparing the above with (11.111), we can conclude that:
C tan _ d
(1 d D )C e _ D (1 d V )C e _ V
wd V 1
wd D 1
dev
dev
sph sph
wt D t D
wt V t V
(11.128)
610
VY
Elastic limit
(1 d ) E
Hp
(11.129)
The next model considers the above plus an additional term which is only used in
terms of the damage variable:
: : e (, d ) : p (B p , d ) : d (B d )
(11.130)
The next model considers energy decomposition into the elastic part : e (, d ) and
: p (B p ) which in turn is only used in terms of the plastic strain:
: : e (, d ) : p (B p )
(11.131)
1 d
(11.132)
: : (, p , q, d ) (1 d ): 0 ( ) : p ;(q, p )
(11.133)
where d is the damage parameter, q is a set of plastic internal variables, p is the plastic
relaxation stress tensor, ;(q, p ) is the plastic potential, and : 0 ( ) is elastic strain
function (energy density). In the particular case when the constitutive relationship is linear,
we have : 0 () : e ()
1
: Ce : .
2
11 DAMAGE MECHANICS
611
Then, taking into account the Helmholtz free energy we can obtain its rate of change as
follows:
w: w: p w:
w:
: (, p , q, d )
:
:
q
d
p
w
wq
(11.135)
wd
q
: p :
w
wq
wd
d t 0
(11.136)
Next, as the above inequality must be satisfied for any admissible thermodynamic process,
we find:
w:
w
(11.137)
and by considering the strain energy equation given in (11.133) we can obtain:
w:
w
(1 d )
w: 0
p
w
w:
w p
w; (q, p ) w:
,
wq
w p
w; (q, p ) w:
,
wd
wq
: 0
(11.138)
w; (q, p ) p w; (q, p )
w: 0
:
q : 0 d t 0
p :
(1 d )
p
w
wq
w
w; (q, p )
w; (q, p )
w: 0
: p
q : 0 d t 0
p :
(1 d )
w
wq
(11.139)
Note that the above inequality must hold for any admissible thermodynamic process, so, let
us consider one in which we have p 0 , q 0 , d 0 , so the only way to fulfill the
entropy inequality in (11.139) is when:
(1 d )
w: 0
p
w
w:
w
(1 d ) 0 p
(11.140)
0 p
(11.141)
Then, by substituting the equation in (11.140) into the inequality (11.139) we find:
w; (q, p )
w; (q, p )
q : 0 d t 0
: p
p
wq
w
(11.142)
Now, we can assume a purely damage process, and another purely plastic process, with
which we obtain the following restrictions:
612
: 0 d t 0
w; (q, p )
w; (q, p )
q t 0
: p
p
wq
w
and
(11.143)
1
2: 0
2
o
t
2: 0 ( ) rate
2:
1
2
( )
1 w: 0
1 0
:
:
t w
t
(11.144)
Note that we are dealing with a symmetrical norm, and also that when the constitutive
: Ce :
2: e which is the same as that
equation is linear we obtain t t
outlined in the isotropic damage model, (see equation (11.31)).
We can describe the damage state in the material by means of the damage criteria in the
strain space by:
G(t , r ) t r d 0
(11.145)
[ H (t , d )
r [ (, r )
(11.146)
G (t , r ) d 0
[ G (t , r )) 0 Kuhn-Tucker condition
(11.147)
(11.148)
w w
d
:
wd
w
(11.149)
w
w
(1 d )
w 2: 0
,
w w
11 DAMAGE MECHANICS
613
undergoing plastic loading we have r [ (, r ) . So, here, the damage consistency parameter
[ ( , r ) can be determined by the persistency condition, [ ( , r ) t
d
H (t , d )
0 : , whereby:
0 :
(11.150)
Then, given all the above considerations, the equation in (11.149) can be rewritten as:
w 2: 0
1 0
: 0
( , d ) (1 d )
: H (t , d )
w w
t
H (t , d ) 0
w 2: 0
0 :
(1 d )
w w
t
(11.151)
H (t , d ) 0
w 2: 0
0
(1 d )
w w
t
(11.152)
Note that the tensor C tan_d features major and minor symmetry.
11.4.1.5 Characterization of the Plastic Response. The ElastoplasticDamage Tangent Stiffness Tensor
Characterization of the plastic response will be formulated in the effective stress space
and p , then the following holds:
( , p )
w: 0 ( )
p
w
(11.153)
We can also postulate the yield function in the effective stress space, so that the elasticdamage domain is characterized by F ( , q) d 0 .
Then, by assuming there is an associated flow rule, the constitutive equations for plastic
response are given by:
w: 0 ( )
wF
p , q
w
Plastic flow rule
H
w
p
q
w: 0 ( )
H h
p , q
w: 0 ( )
p , q d 0
F
w
(11.154)
Yield criterion
where p is the rate of change of the plastic relaxation stress tensor, H is the plastic
consistency parameter, and h is the hardening law and the loading/unloading conditions
are given by:
w: 0 ( )
p , q d 0 ;
H t 0 ;
w: 0 ( )
H F
Kuhn-Tucker
p , q 0
conditions
(11.155)
614
Now, to obtain the value of the plastic consistency parameter H ! 0 we turn to the
F ( , q)
wF wF
>H h , q @ 0
:
wq
w
(11.156)
Note that:
w: 0 ( )
wF
p , q
p
w
H wF ( , ), q : w
p H
w
w
w
The rate of change of is then evaluated as follows:
w: 0 ( )
p
w
H
wF , q w 2 : 0 ( )
:
w
w w
w 2 : 0 ( )
wF
: H
w w
w
w 2 : 0 ( ) p
:
w w
(11.157)
(11.158)
F ( , q)
wF wF
:
q
wq
w
wF
w
w 2 : 0 ( )
w 2 : 0 ( ) wF
:
:
: H
w w w
w w
wF
>H h , q @ 0
wq
wF
wF w 2 : 0 ( ) wF
wF w 2 : 0 ( )
:
: H
:
:
h , q 0
H
wq
w w w w
w w w
wF w 2 : 0 ( ) wF wF
wF w 2 : 0 ( )
:
: H
:
:
h , q
w w w
w
w
q
(11.159)
wF w 2 : 0 ( )
:
:
w w w
2 0
wF w : ( ) wF wF
:
:
h , q
w w w w
wq
(11.160)
w 2 : 0 ( )
wF
: H
w w
w
w 2 : 0 ( ) wF wF w 2 : 0 ( )
:
:
2 0
w : ( ) w w w w w w
:
w w wF w 2 : 0 ( ) wF wF
h , q
:
:
w w w w
wq
(11.161)
Note, the development of the equation (11.161) in indicial notation is similar to that seen in
Chapter 9 in subsection 9.4.4 The Elastoplastic Tangent Tensor.
Then, the effective elastoplastic tangent stiffness tensor is defined as follows:
C tan_p
w 2 : 0 ( ) wF wF w 2 : 0 ( )
:
:
w 2 : 0 ( ) w w w w w w
2 0
w w
wF w : ( ) wF wF
:
:
h , q
w w w w
wq
(11.162)
11 DAMAGE MECHANICS
615
1 d d 1 d C tan_p : d
H (t , d )
(11.163)
obtain:
1 d C tan_p : d 1 d C tan_p : H (t , d ) 0 :
t
(11.164)
Then, by representing the above equation in indicial notation we can conclude that:
V ij
tan_p
1 d C ijkl
H kl
H (t , d )
V 0kl H kl V ij
H (t , d )
tan_p
V ij V 0kl H kl
1 d C ijkl
t
(11.165)
H (t , d )
tan_p
0 :
1 d C
t
C tan_d - p :
(11.166)
tan_d - p
(11.168)
where and are the tensile and compressive effective stress tensors, respectively,
and are defined by means of the spectral representations as follows:
a 1
V a n ( a ) n ( a )
Va
n ( a ) n ( a )
(11.169)
a 1
616
V1
0
Vcij
0
V1 0 0 0 0
0 V 0 0 0
0
2
0
0 0 0 0 V 3
0
0
V 3
0
V2
0
Vij
(11.170)
Vij
:1
Tr ( )
Tr ( )
Tr ( ) Tr ( )
(11.171)
(11.172)
: : , p , d , d 1 d : e 1 d : e
(11.173)
: e (, p )
1
1
: C e : (11.174)
2
where
:e
: e (, p )
1
1
: Ce :
2
:e
1
1
M
1
1 1
I
2N(3M 2N)
2N
1
>(1 Q) I Q1 1@
E
1
1
(1 Q ) E ik E jl E il E jk QE ij E kl
E
2
1
e
e
{ D ijkl
C ijkl
(11.175)
:e
1
1
1
: Ce :
: >(1 Q ) I Q1 1@ :
2
2E
Q
(1 Q)
:
Tr ( ) Tr ( )
2E
2E
(11.176)
Then, taking into account (11.171), the equation in (11.176) can be rewritten as follows:
:e
Q
(1 Q)
Tr ( ) Tr ( )
:
2E
2E
(1 Q)
Q
Tr ( ) Tr ( ) Tr ( )
:
2E
2E
2
(1 Q)
(1 Q)
Q
Q
Tr ( )
Tr ( ) Tr ( )
:
:
E
E
2E
2E
2
2
0
>
>
(11.177)
:e
>
2
(1 Q)
Q
Q
:
Tr ( )
Tr ( ) Tr ( )
2E
2E
2E
1
(1 Q)
1
Q
Q
:
Tr ( ) Tr ( )
: Ce :
Tr ( ) Tr ( )
2E
2E
2
2E
(11.178)
11 DAMAGE MECHANICS
617
:e
1
1
1
1
1
1
1
1
: Ce :
: Ce :
: Ce : : Ce :
2
2
2
2
1
1
1
: >(1 Q) I Q1 1@: : C e :
(11.179)
2E
2
1
Q
(1 Q )
1
Tr ( ) Tr ( ) : C e :
:
2E
2
E
2
0
1
Q
1
Tr ( ) Tr ( ) : C e : t 0
E
2
2
t0
(11.180)
t0
Now, the damage variable values lie between the following ranges:
0 d d d1
0 d d d1
(11.181)
w: w:
w:
w:
:
: p d d
p
w
w
wd
wd
(11.182)
Remember from subsection 6.4.1 Constitutive Equations with Internal Variables in Chapter 6,
w:
that the terms
A i were denoted by the thermodynamic forces, where B i are the
wB i
set of internal variables. Therefore, with the denotations B 1 d , B 2
A 2 A , the thermodynamic forces becomes:
A
w:
wd
:e
A
w:
wd
:e
d , A1
A and
(11.183)
1 d :
e
1 d
e
w:
wd
w:
wd
: e
: e
(11.184)
: Ce
1
:
(11.185)
In compression, the norm (based on the Drucker-Prager model and which was obtained by
Faria&Oliver (1993)) can be expressed in terms of the normal octahedral effective stress
) and is given by:
( V oct ) and octahedral tangential stress ( W oct
618
q (V )
3 K V oct
W oct
(11.186)
proof of which can be found in the chapter on Plasticity (the Drucker-Prager Criterion).
We will now introduce two damage criteria denoted by H and H :
H (t , r ) t
r d 0
H (t , r ) t
r d 0
(11.187)
where r and r are the current damage thresholds which serve to remind us where the
damage surface is during the loading/unloading/loading process and whose initial values
are represented by r0 and r0 , respectively.
Then, by defining f 01D and f 01D as the stresses after a visible non-linearity in a uniaxial
1
Tr ( )
3
1
D
f 0 , W1oct
3 1D
2
J2
3
2
f0 ,
3 1D
(see Chapter 9 in the subsection: 9.2.4.2 The Drucker-Prager Yield Criterion), from which
we define the following elastic thresholds:
r0
f 0
1D
f 0
1
f0
E 1D
1D
3
( K 2 ) f 0
1D
3
r0
(11.188)
Tension:
d
Compression:
d
w) (t )
[
wt
(t )
)
w
[
wt
[
r
(11.189)
[
r
(11.190)
H d 0
[ H
(11.191)
[ t 0
H d 0
[ H
(11.192)
Compression
We can verify that when H 0 damage ceases and from the Kuhn-Tucker conditions we
obtain [ 0 . When [ ! 0 there is damage evolution and here the Kuhn-Tucker
conditions hold if H 0 , that is, providing that the current state is on the damage surface.
In these conditions it is also true that t r and:
r
likewise, for r :
(11.193)
11 DAMAGE MECHANICS
r
619
(11.194)
t
w) (t )
wt
t0
)
d
t
w) (t )
wt
t0
)
(11.195)
p
(11.196)
where E is the Youngs modulus and E t 0 is also a material parameter that will control
the rate intensity of plastic strain. The value E 0 is equivalent to the elastic damage case.
H (d ) is the Heaviside function of the compression damage rate, which was introduced to
(11.197)
: e 1
EEH (d )
C :
:
EEH (d )
:
EEH (d )
1
: C e :
1
: :
Ce :
1
EEH (d ) : C e :
(11.198)
wd
wd
(11.200)
and by substituting this into the internal energy dissipation given in (11.199), we find:
w: w: p w: w:
: p : d d t 0
w
wd
wd
w
(11.201)
Then, as the above inequality must be satisfied for any admissible thermodynamic process,
we can draw the conclusion that:
w:
w
(11.202)
620
In a small deformation regime, we can decompose additively the infinitesimal strain tensor
into elastic ( e ) and plastic ( p ) parts, i.e.:
e p
p
(11.203)
: : e , d , d 1 d : e ( e ) 1 d : e ( e )
(11.204)
w:
w
w e
w
equation becomes:
1 d w:
w
1 d : e ( e ) 1 d : e ( e )
w
e
e
e
e
(
)
(
:
:
e ) w e
w
w
w
1 d
:
1
d
:
w
w
w e
w e
e
( e )
w e
1 d
w:
e
(11.205)
1 d 1 d
( e )
w e
(11.206)
w: e ( e )
w e
1 d
w: e ( e )
w e
1 d
(11.207)
1 d :
e
1 d : e
1 d 12
1
1
1
1
: Ce : : Ce :
2
2
1
1
1
: Ce :
:
2
2
: Ce
1
: 1 d
12
: Ce
1
1
1
: Ce :
2
(11.208)
where we have taken into account the expressions of : e and : e given in (11.174).
Then, if we consider the internal energy dissipation in (11.201) and the constitutive
equation in (11.202), we can conclude that:
Dint
w: p w: w:
: d d
w p
wd
wd
w: p
: : e d : e d t 0
w p
(11.209)
d t 0
(11.210)
If we now consider there to be a purely plastic process, the internal energy dissipation
becomes:
11 DAMAGE MECHANICS
w: p
:
w p
w: ( e ) w e p
: p :
w e
w
w 1
:
w e 2
w: ( e )
: I : p
w e
621
w: ( e ) p
: t 0
w e
(11.211)
w: ( e )
can be obtained directly from the equation in (11.208):
w e
w
1 w
: : e
2 w e
w
1 e
C : :I
2
1
2
(11.212)
1
aC e :
with
:
t0
a EEH (d )
:
(11.213)
and by incorporating the equations (11.212) and (11.213) into the internal energy
dissipation in (11.211), we can obtain:
w: ( e ) p
: t 0
w e
1
a : Ce : t 0
2 a: t 0
(11.214)
(11.215)
Further details about the numerical implementation of the tensile-compressive plasticdamage model are described in Faria&Oliver (1993).
622
E
2
C
A
B
H (1)
H (2)
(11.216)
0d s dt
In this model Gurtin&Francis adopted as the constitutive equation for stress, V , by means
of the current strain state and the damage variable as follows:
f ( ] ) g (H m )
(11.217)
H
Hm
(11.218)
The function f (] ) , called the damage master curve, defines the loss of stiffness and satisfies:
(11.219)
f (1) 1
Then, when the maximum strain takes place in the current time (H m
is given by:
V
g (H m )
(11.220)
Then, the function g defines the uniaxial stress-strain curve obtained from a
monotonically increasing/decreasing uniaxial test. In Figure 11.14, this function is defined
according to path ACE .
Now, to fully describe the material parameters in this model, we need to determine both
the virgin curve and the damage master curve f (] ) . This latter curve is obtained from the
uniaxial test.
11 DAMAGE MECHANICS
623
Let us now consider a isotropic hyperelastic material, (see Chapter 8), which is governed by
the free energy, : 0 , described in terms of the principal stretches ( O 1 , O 2 , O 3 ). Now, the
Kirchhoff stress tensor eigenvalues, in terms of principal stretches, can be expressed as
follows:
Wa
w: 0
: g a (O 1 , O 2 , O 3 )
wO a
Oa
(11.221)
The above equation is valid only when we are dealing with virgin material during loading.
Then, the general form of (11.221) can be expressed as follows:
Wa
f ( [) g a ( O 1 , O 2 , O 3 )
(11.222)
As when we looked at the 1D case, we will define a function dependent on the relative
strain, [ , in 3D, in which the following remains valid:
f (1) 1
(11.223)
max : 0 ( s )
d (t )
0d s dt
(11.224)
:0
(11.225)
If we draw an analogy with the yield surface from classical plasticity, we can define a damage
surface in the principal stretch space as follows:
) (O 1 , O 2 , O 3 , d ) : : 0 (O 1 , O 2 , O 3 ) d
(11.226)
For a fixed value of d , the damage surface limits a region in the principal stretches space
where the material behavior is purely elastic, i.e. where there is no damage evolution. As
with plasticity, the damage variable evolution is characterized by the loading/unloading
condition:
)d0
d t 0
d)
(11.227)
Damage Variable
d (t ) max : 0 ( s )
ii.
0 d s dt
[:
:0
(11.228)
d
iii. Damage Surface
) (O 1 , O 2 , O 3 , d ) : : 0 (O 1 , O 2 , O 3 ) d
iv. Loading/Unloading Criterion
)d0
d t 0
d) 0
624
(1 d )Z 0 (C )
S0
E : S p ;( A, S p )
1 1
C 1 : S p ;( A, S p )
(1 d )Z (C )
S 0 2
(11.229)
where C
S0
unit mass.
11.6.4.2 Internal Energy Dissipation. Constitutive
Equations. Thermodynamic Considerations
Remember that in Chapter 5 the alternative form of the Clausius-Planck inequality in the
reference configuration is given by:
>
Dint S : E S 0 IT Z t 0
(11.230)
Dint S : E S 0 Z t 0
(11.231)
in which all variables are described in the reference configuration, and S 0 is the mass
density, I is the specific entropy, T is temperature, and Z is the specific Helmholtz free
energy (per mass unit). In isothermal processes we have T 0 , in which the Clausius-Planck
inequality becomes:
Then, by evaluating the rate of change of the Helmholtz free energy, Z(C , S p , A, d ) , we
obtain:
wZ wZ p wZ wZ
Z (C , S p , A, d )
:C
:S
A
d
p
wC
wA
wS
(11.232)
wd
Next, if we consider the expression of the Helmholtz free energy given in (11.229) we find:
wZ
wC
wZ
wd
(1 d )
wZ
1
wZ 0 (C )
Sp,
wC
2S 0
wS p
Z (C )
0
S0
E
w;( A, S p ) wZ
,
wA
wS p
w; ( A, S p )
,
wA
(11.233)
11 DAMAGE MECHANICS
625
After that, substituting (11.232) into the entropy inequality (11.231), and by considering the
equations in (11.233), we have:
w;
wZ 0 (C )
1
1 p w;
S : E S 0 (1 d )
S p : C p
E :S
A Z 0 (C )d t 0
2S 0
wA
wC
S0
wS
(11.234)
or:
wZ (C )
S : E S 0 (1 d )
:C
wC
0
w;
w;
1 p
1 p
S :C S0 p
E :S S0
A S 0 Z 0 (C )d t 0
S
wA
2
w
S
0
1
C 1 E
2
wZ 0 (C )
wC
(11.235)
wZ 0 ( E )
, and in addition
wE
1
C holds, so we can now rewrite the entropy inequality as:
2
wZ 0 ( E )
w;
w;
A S 0 Z 0 (C )d t 0
E : S p S 0
S p : E S 0
S S 0 (1 d )
p
wA
wE
wS
(11.236)
which must be satisfied for any admissible thermodynamic process, so, by considering the
following process S p 0 , A 0 , d 0 , we obtain:
wZ 0 ( E )
S p : E t 0
S (1 d )S 0
w
E
(11.237)
(1 d )S 0
wZ 0 ( E )
Sp
wE
(1 d )S 0 S p
(11.238)
wZ 0 ( E )
is the non-damage stress tensor. Additionally, if we take into
wE
account that S (1 d )S and S p (1 d )S p , where S p is the plastic relaxation effective
where S 0
S0
(1 d )S 0 S p
(1 d )S 0 (1 d )S p
0
S S
(11.239)
Then, substituting the constitutive equation in (11.238) into the inequality in (11.236)
yields:
w;
w;
A S 0 Z 0 (C )d t 0
E : S p S 0
S 0
p
wA
wS
(11.240)
Similarly, if there is a pure damage process and then another pure plastic process, we can
obtain the following restrictions:
626
S 0 Z 0 (C )d t 0
w; ( A, S p )
w; ( A, S p )
S 0
E : S p S 0
At0
p
wA
wS
(11.241)
tE
1
2: 0
2
2: 0 ( E ) rate
o
t E
2:
1
2
(E)
1 w: 0
:E
t E wE
tE
S 0 : E
(11.242)
We emphasize here that : 0 ( E ) S 0 Z 0 ( E ) has the unit of energy per unit volume (energy
density).
We can then characterize the state of damage in the material by means of the damage
criterion defined in the principal strain space as:
G(t E , r ) t E r d 0
(11.243)
[ H (t E , d )
r [ ( E , r )
(11.244)
where [ is the damage consistency parameter which defines the loading/unloading KuhnTucker conditions:
r [ t 0
G (t E , r ) d 0
[ G (t E , r ) 0 Kuhn-Tucker conditions
(11.245)
(11.246)
wS wS wS wS
d
d{
:E
:E
wd
wd
wE
wE
wS wS
d , d [ H (t E , d ) , (see
:E
wd
wE
equation (11.244)), and also that in plastic loading r [ ( E , r ) holds, in which the damage
wS
wd
(11.247)
S 0 , S ( E , d )
tE
w 2 Z 0 ( E ) wS
S ( E , d ) (1 d )S 0
[ H (t E , d )
:E
wE wE
wd
w 2Z 0 (E)
1 0
(1 d )S 0
: E S0
S : E H (t E , d )
wE wE
tE
11 DAMAGE MECHANICS
S ( E , d )
627
w 2 Z 0 ( E ) H (t E , d ) 0
S S 0 : E
(1 d )S 0
w
w
t
E
E
E
(11.248)
Thus, we can define the hyperelastic-damage tangent stiffness tensor in the reference
configuration as:
C tan_d
(1 d )S 0
w 2 Z 0 ( E ) H (t E , d ) 0
S S0
tE
wE wE
(11.249)
The tensor C tan_d features major and minor symmetry, which is, primarily, due to the fact
that the norm adopted is symmetrical.
11.6.4.5 Characterization of the Plastic Response. The
Elastoplastic-Damage Tangent Stiffness Tensor
Effective
Characterization of the plastic response is formulated in the effective stress spaces S and
S p after which the following holds:
S
S0
wZ 0 ( E)
Sp
wE
(11.250)
We will now postulate the yield function in the effective stress space, such that the elasticdamage domain is characterized by F ( E , S p , A) d 0 .
Then, if we assume the associated flow rule holds, the constitutive equations for the plastic
response are given by:
Sp
H
wF ( E , S p , A)
wE
A H H ( E , S p , A)
F ( E , S p , A) d 0
(11.251)
where S p is the rate of change of the plastic relaxation effective stress tensor, H is the
plastic consistency parameter, and H is the hardening law.
F ( E , S p , A) d 0 ;
H t 0 ;
H F ( E , S p , A) 0 Kuhn-Tucker conditions
(11.252)
Then, to obtain the value of the plastic consistency parameter H ! 0 we turn to the
F ( E , S p , A)
wF wF p wF
A 0
:E
:S
wE
wA
wS p
wF wF wF ( E , S p , A) wF
:E
: H
H H S , A
wE
wE
wS p
wA
> @
(11.253)
0
wF
:E
wE
wF wF wF
:
H S, A
wS p wE wA
(11.254)
628
w: 0 ( E )
Sp
wE
S
w 2: 0 (E ) p
:E S
wE wE
(11.255)
w 2: 0 (E ) p w 2: 0 (E)
wF
:E S
: E H
wE wE
wE wE
wE
wF
:E
w 2: 0 (E )
wE
wF
:E
wF wF wF
wE wE
H wE
:
p
w
w
E
A
w
S
(11.256)
wF
wF
w 2: 0 (E)
w
w
E
E
: E
wF wF wF
wE wE
H
:
wS p wE wA
Next, we will define the effective elastoplastic tangent stiffness tensor as follows:
C
tan_p
wF
wF
w 2: 0 (E )
w
wE
E
wF wF wF
wE wE
:
H
wS p wE wA
(11.257)
(1 d )S d S
1 d C tan_p : E d S
H (t E , d )
tE
(11.258)
S 0 : E , we obtain:
1 d C tan_p : E d S 1 d C tan_p : E H (t E , d ) S 0 : E S
tE
H (t E , d )
tan_p
S S 0 : E
1 d C
tE
(11.259)
or:
S
C tan_d - p : E
(11.260)
11 DAMAGE MECHANICS
629
>
: : ( E , E p , A, d ) (1 d ): 0 ( E , E p , A) (1 d ) : e0 ( E , E p ) : 0p ( A)
(11.262)
1
C 1 is the Green-Lagrange strain tensor, C F F T is the right Cauchy2
Green deformation tensor, F is the deformation gradient, and E p is the plastic part in the
where E
reference configuration.
>
Dint S : E S 0 IT : t 0
(11.263)
where S 0 is the mass density, I is the specific entropy (per unit mass), T is temperature,
and : is the Helmholtz free energy (per unit volume). In isothermal processes we have
T 0 , with which the Clausius-Planck inequality becomes:
Dint S : E : t 0
(11.264)
:
(1 d )
w:
e
p
0 (E, E )
wE
wA
wE
: E (1 d )
w:
wd
e
p
0 (E, E )
p
wE
: E p
w: 0p ( A)
A : 0 ( E , E p , A)d
wA
(11.265)
Next, substituting the above equation into the internal energy dissipation given in (11.264),
yields:
S : E (1 d )
w: 0e ( E , E p )
w: 0e ( E , E p ) p w: 0p ( A) w: 0 ( E , E p , A)
d t0
A
: E (1 d )
:E
wE
wA
wd
wE p
w: 0e ( E , E p )
w: 0p ( A)
w: 0e ( E , E p )
: E (1 d )
S (1 d )
: E p
p
wE
wA
wE
A : 0 ( E , E p , A)d t 0
(11.266)
Then, as the above inequality must satisfy for any admissible thermodynamic process, we
obtain:
S
(1 d )
w: e0 ( E , E p )
wE
(11.267)
630
(1 d )
w: 0e ( E , E p )
wE
: E p (1 d )
w: 0p ( A)
A : 0 ( E , E p , A)d t 0
wA
(11.268)
(11.269)
w: 0e ( E , E p )
wE
w: 0 ( A)
: E p
At0
wA
p
(11.270)
1
S
(1 d )
w: e0 ( E , E p )
wE
11.6.5.3 Characterization
of
Hyperelasticity Tensor
(11.271)
Damage.
The
Tangent
Damage
(11.272)
(11.273)
where ([ t , r t ) represent the current values of ([, r ) , at time t , where r t is the current
damage threshold, and by adopting r0 as the material elastic threshold r t t r0 holds.
The evolution of the variables d and r are then given, respectively, by:
d [ ; (d t , [ t , s, a) ;
r t [
(11.274)
H([ t , r t ) d 0
(11.275)
[
(11.276)
[
rt
max r0 , max[ t
s>0,t @
(11.277)
11 DAMAGE MECHANICS
631
w 2 : e0 ( E , E p ) w: e0 ( E , E p )
S ( E , d ) (1 d )
:E
d
wE wE
wE
w 2 : 0e ( E , E p )
w 2 : 0e ( E , E p )
: E S d { (1 d )
: E S d
S ( E , d ) (1 d )
wE wE
wE wE
w 2 : 0e ( E , E p )
: E ; S [
(1 d )
wE wE
(11.278)
Note that during loading d r; [ ; is valid, and also based on the expression
[ { : 0 ( E , E p , A) , the non-plastic process, we obtain:
[
w: 0
:E
wE
S : E
(11.279)
Then
w 2 : 0e ( E , E p )
w 2 : 0e ( E , E p )
S ( E , d ) (1 d )
: E ; S [ (1 d )
: E ; S S : E
wE wE
wE wE
(11.280)
w 2 : 0e ( E , E p )
(
1
)
S
S
:
;
d
E
wE wE
(1 d )
w 2 : 0e ( E , E p )
;S S
wE wE
(11.281)
Response.
The
elastoplastic
Now, by assuming there is an associated flow rule, the constitutive equation for the plastic
w 2 : e0 ( E , E p )
wS
wF ( E , E p , A)
response is given by E p H M 1 :
, where M
.
p
p
wE
wE wE
wE
M : E p
H
wF ( E , E p , A)
wE
(11.282)
(11.283)
F ( E , E p , A) d 0 ;
H t 0 ;
H F ( E , E p , A) 0 Kuhn-Tucker conditions
(11.284)
F ( E , E p , A)
wF wF p wF
A 0 , which can be expressed as:
:E
:E
wE
wA
wE p
wF wF
wF wF
>H H @ 0
:E
: H M 1 :
wE
wE wA
wE p
(11.285)
632
wF
:E
wE
wF
wF wF
: M 1 :
H
p
wE wA
wE
w: e0 ( E , E p )
is evaluated as:
wE
w 2 : 0e ( E , E p ) w 2 : 0e ( E , E p ) p w 2 : 0e ( E , E p )
:E
:E
: E M : E p
wE wE
wE wE
wE wE p
w 2 : 0e ( E , E p ) p w 2 : 0e ( E , E p )
wF
:E S
: E H
wE wE
wE wE
wE
(11.286)
(11.287)
w 2 : e0 ( E , E p )
, and that the value of H is given by the
wE wE
w 2 : e0 ( E , E p )
wF
wF
: E H
C e0 : E H
wE wE
wE
wE
wF
wF
wF
:E
wF e
e
w
E
E
E
w
w
C0 : E
C 0
: E
wF
w
w
F
F
F
F
F
w
w
w
wE
1
1
:M :
:M :
H
H
wE wA
wE wA
wE p
wE p
(11.288)
thus we can define the effective elastoplastic tangent stiffness tensor ( C
tan _ p
C 0e
tan _ p
) as:
wF
wF
w
E
wE
wF
wF
1 wF
:M :
H
wE wA
wE p
(11.289)
(1 d )S d S
(1 d ) C tan _ p : E [ ; S
(11.290)
wE
[ { : 0 ( E , E p , A)
or
(11.291)
11 DAMAGE MECHANICS
633
wF w: 0
w:
[ S : E H 0p : M 1 :
H
wA
wE
wE
wF
:E
w
E
w: 0 : M 1 : wF w: 0 H
S:E
w
wF
w
p
F
F
wE
wA
: M 1 :
H wE
p
wE wA
wE
wF
:E
>Y @
S : E wE
X
Y
S
w: 0
wE p
: M 1 :
(1 d ) C
(11.292)
wF
wF wF
: M 1 :
H,
wE wA
wE p
wF w: 0
H . Then, substituting (11.292) into (11.290) yields:
wE
wA
tan _ p
: E ;[ S
(1 d ) C
tan _ p
Y
(1 d ) C tan _ p : E ; S : E S ;
X
wF
:E
>Y
: E ; S : E wE
X
@ S
(11.293)
wF
:E S
wE
in indicial notation:
S ij
Y
tan _ p
(1 d ) C ijkl
E kl ; S kl E kl S ij ;
X
Y
tan _ p
(1 d ) C ijkl ; S kl S ij ;
X
wF
E kl S ij
wE kl
wF
S ij E kl
wE kl
(11.294)
Y
tan _ p
; S S ;
(1 d ) C
X
wF
S
:E
wE
wF w: 0
w: 0
H
: M 1 :
p
wF
E
A
w
w
tan _ p
E
w
:E
(1 d ) C
S
;S S ;
wF
wF
wE
1 wF
M
H
:
:
wE wA
wE
tan _ d _ p
C
:E
(11.295)
C tan _ d _ p
(1 d ) C tan _ p
wF w: 0
w: 0
: M 1 :
H
p
w
w
E
A
S wF
; S S ; wE
wF
w
w
F
F
wE
: M 1 :
H
p
w
w
E
A
w
E
(11.296)
12 Introduction to Fluids
12
Introduction to Fluids
12.1 Introduction
In this chapter, we will introduce an important branch of continuum mechanics: fluid
mechanics with which we intend to study fluids in motion or at rest. These can be classified
into:
Liquids
Fluids
Gases
There are several areas where fluids mechanics can be applied, e.g. meteorology,
oceanography, aerodynamics, hydrodynamics and engineering, among others.
Fundamentally, we can state that solids can resist shear stress while liquids have very low
(viscous fluids, e.g. oil) or no resistance to it (non-viscous fluids, e.g. water).
Both gases and liquids are materials consisting of molecules (an agglomeration of two or
more atoms) colliding with each other. To treat fluids with assumption of continuum
mechanics properties (e.g. mass density, pressure and velocity) are treated as continuous
functions. Then, treating a system of molecules as a continuous medium is valid when
comparing the mean free path of molecules ( / ) (average distance particles travel before
colliding with each other) with the characteristic physical length scale ( " C ).For example,
for solids and liquids we have / | 10 7 cm and for gases / | 10 6 cm , Chung (1996). Then,
the ratio
"C
unity, the domain can be treated as a continuum, otherwise we must use statistical
mechanics to obtain the governing equations of the problem with which we can establish
that:
635
636
Kn
Kn
"C
1
"C
macroscopic approach
(12.1)
!1
microscopic approach
Let us now consider a fluid is between two surfaces separated by a distance d , (see Figure
12.1). The lower surface is fixed whilst the upper surface is moving at the constant velocity
v 0 . We can then observe that the force required to maintain this motion is given by:
F
A
v0
d
(12.2)
where A is the surface area and N is the fluid viscosity. The above equation indicates that
the shear stress
F
is proportional to how the velocity varies with distance (the velocity
A
gradient).
in motion
A
v v0
d
v 0
fixed
Figure 12.1: Motion of the plate.
components
t i(n) p 0 n i
(12.3)
p
V kk
3
(12.4)
12 INTRODUCTION TO FLUIDS
637
In this case, any direction is a principal direction and the stress tensor, ( p 0 ) , is
represented as follows:
Tensorial notation
Components
V ij p 0 E ji
p0 1
(12.5)
The constitutive law above was described by Bernoulli for a non-viscous fluid.
Unfortunately this equation is not valid for any fluid, for instance a fluid in motion. The
negative sign indicates a compressive stress with a positive pressure value.
Figure 12.2 shows the hydrostatic state by means of the Mohrs circle, here reduced to a
single point. Note that for a fluid at rest the maximum tangential (shear) stress is zero and
here the stress tensor is spherical.
VS
p0
p0
V1 V 2 V3
p0
VN
b) Mohrs circle
p1
V ij
pE ji W ij
(12.6)
638
and non-viscous fluids, e.g. water (an incompressible non-viscous fluid), air (a compressible
non-viscous fluid) and oil (incompressible viscous fluid).
|0
Non-viscous fluids
(12.7)
In this case we have a perfect fluid and the equation in (12.3) holds.
pE ij E ij W ij E ij
V ii
3 p W ii
1
V ii
3
,
: 1 p1 : 1 : 1
Tr ( ) 3 p Tr ( )
1
p W ii
3
Tr ( )
3
p
Tr ( )
3
(12.8)
Vm
p0
0 , thus:
(12.9)
Compressible Fluids
In general, pressure is a function of mass density and temperature which are
related by means of the equation of state:
p
p (S , T )
(12.10)
where p is pressure, S is mass density, and T is absolute temperature. For example, the
equation of state for ideal gases is p SRT , where R is the gas constant.
When temperature is not included in the equation of state, S S ( p ) , then the state change
is called barotropic.
It can then be shown that in reversible adiabatic processes, an ideal gas is governed by the
barotropic relationship:
p
constant
cp
cv
1
R
cv
(12.11)
where c p is specific heat at constant pressure, c v is the specific heat at a constant volume,
R is the gas constant and a perfect incompressible fluid is governed y the barotropic
equation:
S constant
(12.12)
12 INTRODUCTION TO FLUIDS
where we have p
639
p0 .
&
v "C
N*
&
v "C
(12.13)
Q*
&
where v is the mean velocity of the object relative to the fluid; " C is the characteristic
length; N * is the dynamic viscosity, and Q * is the kinematic viscosity which is given by
N*
kg m 2 kg
. Then, the SI units of these are Q *
m 2 / s and N *
Pa u s .
Q*
ms
S
m3 s
> @
> @
Laminar flow
Turbulent flow
640
& &
& &
& &
Dv ( x , t ) & &
wv ( x , t )
{ v ( x, t )
x& v v
Dt
wt
&
(12.14)
Dvi
wvi ( x , t )
&
ai ( x, t )
{ vi
v k vi ,k
Dt
wt
&
The spatial velocity gradient ( x& v { l ) can be split into a symmetric and an antisymmetric
& &
a ( x, t )
& &
& &
a ( x , t ) v ( x , t )
(12.15)
&
&
&
1
l l
2
&
Dv
v&
& &
&
( sym
x v) v
D ij v j
1
v i , j v j ,i v j
2
(12.16)
& &
& &
&
& wv ( x , t )
& & 1 & & &
wv ( x , t )
&
x v v
Dv W v
( sym
x v) v
wt
wt
2
& &
& & 1 & &
wv ( x , t )
&
( sym
v
x v) v
wt
2
& &
a ( x, t )
(12.17)
Finally, remember in Chapter 2 it was shown that the following equation is valid:
& &
a ( x, t )
& &
& &
wv ( x , t ) 1 & 2
x (v ) rot v v
wt
2
& &
& &
wv ( x , t ) 1 & 2
x (v ) x& v v
wt
2
(12.18)
Tr (D) 0
DS
Dt
0
o S
S 0 Incompressible fluids
(12.19)
12 INTRODUCTION TO FLUIDS
641
Wij
0 ij
&
x& v
>
&T
1 &&
x v x& v
2
Irrotational flow
&
0
(12.20)
&
x& v
and
&
0
(12.21)
& &
& & 1 & &
wv ( x , t )
&
v
( sym
x v) v
2
wt
&
0
& &
& &
wv ( x , t )
&
( sym
x v) v
wt
Note that v k v k ,i
&
wvi ( x , t )
sym
v k vi ,k
wt
&
ai ( x, t )
(12.22)
following is valid 2v k ,i v k
v k v k ,i , i.e.
&
&
Dv
vi ,k , the
1 & 2
x (v ) , with which
2
& &
& &
wv ( x , t )
&
( sym
x v) v
wt
& &
wv ( x , t ) 1 & 2 Acceleration for an
x (v )
irrotational flow
wt
2
(12.23)
&
0 Steady flow
(12.24)
Remember that the material time derivative of the velocity is given by:
&
Dv &
{v
Dt
& &
& &
wv ( x , t )
x& v v
wt
Dvi
{ vi
Dt
&
wv i ( x , t )
v i ,k v k
wt
(12.25)
& &
x& v v
vi
(12.26)
Problem 12.1: Demonstrate whether the following statements are true or false:
a) If the velocity field is steady, then the acceleration field is also;
b) If the velocity field is uniform, the acceleration field is always equal to zero;
c) If the velocity field is steady and the medium is incompressible, the acceleration is always
zero.
Solution:
642
& &
wv ( x , t) &
0 whereby the acceleration field becomes:
wt
&
wvi ( x , t )
vi ,k v k v i ,k v k
wt
vi
0i
&
a
&
v
& &
& & & &
& & & &
wv ( x )
x& v ( x ) v ( x ) x& v ( x ) v ( x )
wt
Independen t of time
&
v
& &
& &
& &
wv ( x , t )
x& v ( x , t ) v ( x, t )
&
wt
0
& &
wv ( x , t )
wt
&
v
& &
& & & &
& & & &
wv ( x )
x& v ( x ) v ( x ) x& v ( x ) v ( x )
wt
(D)
(12.27)
where D is the symmetric part of the spatial velocity gradient (see Chapter 2), whose
components are given by:
D ij
1 wv i wv j
2 wx j wx i
(12.28)
The equation in (12.27) is, general speaking, nonlinear, which is characteristic of the
Stokesian (non-Newtonian) fluid. Blood, paints and sauces are all examples of this.
When the relationship in (12.27) is linear we have what we term Newtonian fluids which are
described in the following format:
W ij
K ijkl D kl
K : (D)
(12.29)
p1
V ij
pE ji W ij
(12.30)
12 INTRODUCTION TO FLUIDS
643
Then, to directly obtain an expression for , we can make an analogy with the stress
constitutive equation for isotropic solid materials, (see Chapter 7), in which:
Isotropic solids
Fluids
V ij
V ij
C ijkl H kl
ME ij H kk 2NH ij
W ij
W ij
K ijkl D kl
M *E ij D kk 2N *D ij
(12.31)
where M * is the viscous dilatational coefficient and N * is the viscous tangential coefficient
and, generally speaking, these variables are associated with other thermodynamics variables,
i.e.:
M * M * (S , T )
N * N * (S , T )
(12.32)
Now, by substituting the viscous stress tensor given in (12.31) into the equation in (12.30)
we obtain:
V ij
pE ij M *D kk E ij 2N *D ij
p1 M * Tr (D)1 2N *D
Navier-Poisson law
(Newtonian fluid)
(12.33)
p1 2N *D
Navier-Poisson law
(Incompressible Newtonian fluid)
(12.34)
pE ij E ij M *E ij E ij D kk 2N *E ij D ij
V kk
pE kk M * E kk D ii 2N *D kk
V kk
3 p 3M *D kk 2N *D kk
1
V kk
3
1
V kk
3
(12.35)
p M * N * D kk
3
p k *D kk
p p
k *D kk
which thus defines the bulk viscosity coefficient (also called the volume or second viscosity)
( N * ) as:
N*
M*
2N *
3
(12.36)
dev
3 1
V
V
V ijdev kk E ij
ij
3
Tr (D)
dev
D D 3 1
D
D
D ijdev kk E ij
ij
3
(12.37)
644
pE ij M * E ij D kk 2N *D ij
V ij
V ijdev
V kk
E ij
3
,
pE ij M * E ij D kk 2N * D ijdev kk G ij
3
p
(12.38)
2N *
D kk E ij 2N *D ijdev
pE ij M *
3
Related to
shape change
p p 1 N * Tr(D)1 2N *D dev
p p E ij N * D kk E ij 2N *D ijdev
(12.39)
The above can be decomposed into two sets of equations. Then, we can consider the
equations in (12.35) in which p p N *D kk holds, which, substituted into the equation
in (12.39), yields:
dev
2N *D dev
2N *D ijdev
V ijdev
(12.40)
Then, the equations in (12.39) can be replaced with the following set of equations:
dev 2N *D dev
p p N * Tr (D)
(12.41)
Tr (D W )
(12.42)
Tr (D) D kk
where l is the spatial velocity gradient, and the trace of D can be expressed in terms of
velocity divergence by:
D kk
wv1 wv 2 wv3
wx1 wx 2 wx3
&
x& v
(12.43)
and by considering the mass continuity equation, (see Chapter 5), the following remains
valid:
DS
&
S x& v
Dt
&
0 x& v
p N*
1 DS
D kk
S Dt
(12.44)
p N * Tr (D) we obtain:
1 DS
S Dt
(12.45)
0 i.e.
DS
Dt
0 (Incompressible fluid)
(12.46)
0 (Stokes condition)
(12.47)
12 INTRODUCTION TO FLUIDS
645
dev
2N *D dev
(12.48)
O*
2N *
3
(12.49)
This condition ensures us that the pressure p is defined as the average of the normal
stresses, i.e.:
1
V ii p N * Tr (D)
o V m
3
,
p
o p
(12.50)
Vm
V ij D ij
pE ij D ij M *E ij D kk D ij 2N *D ij D ij
pD ii M *D kk D ii 2N *D ij D ij
(12.51)
Then, by splitting the rate-of-deformation tensor into a deviatoric and spherical part,
( D ij D ijdev 13 D kk E ij ), and by substituting them into the equation in (12.51) we can obtain:
:D
2
1 : D
1
pTr (D) M * >Tr (D)@ 2N * D dev
3
3
(
D
)
>
@
Tr
2
pTr (D) M * >Tr (D)@ 2N *
D dev : D dev
(12.52)
2N *
>Tr (D)@2 2N *D dev : D dev
pTr (D) M *
or in indicial notation:
V ij D ij
D pp
E ij
pD ii M *D kk D ii 2N * D ijdev kk E ij D ijdev
3
3
D
D
kk pp
pD ii M *D kk D ii 2N *
D ijdev D ijdev
3
(12.53)
646
where the deviatoric tensor trace is equal to zero, D dev : 1 0 . Then, by restructuring the
above equation we obtain:
V ij D ij
2N *
D kk D ii 2N *D ijdev D ijdev
pD ii M *
3
pD ii N D kk D ii 2N
*
(12.54)
D ijdev D ijdev
(12.55)
2WD
The term ( pTr (D) ) is related to elastic energy, so, it is recoverable and because of this we
have the definition:
pTr (D)
Recoverable power
(12.56)
which does not contribute to internal entropy generation in the system. We can also define
the dissipated power per unit volume ( 2W D ), which is associated with internal energy
dissipation:
2W D
2W D
N *D kk D ii 2N *D ijdev D ijdev
N * >Tr (D)@ 2N *D dev : D dev
2
Dissipated power
(12.57)
NOTE: As we can verify by looking at the dissipated power, all dissipated energy is
brought about by viscosity when the fluid is in relative motion between particles, (see
Figure 12.5).
Dissipated energy
viscous fluid in motion
particle
Figure 12.5: Viscous fluid in motion.
Now, by considering the second law of thermodynamics (nonnegative dissipation), we can
conclude that:
( N* ! 0 )
and
N* t 0 M * t
2N *
3
(12.58)
0 , and the stress
(12.59)
12 INTRODUCTION TO FLUIDS
647
&
DS
S ( x& v ) 0
Dt
&
&
x& Sb Uv
S u
DS
Svi ,i
Dt
;
;
V ij , j S b i
&
: D x& q Sr
u
(12.60)
Svi
(12.61)
V ij D ij
or
q i ,i r
&
(12.62)
2WD
&
where u is the specific internal energy, r is the heat generated by internal sources and q is
the heat flux vector (non-convective).
The mass continuity equation, the equations of motion and the energy equation give us five
&
equations in total. The unknowns are: velocity v (three components), temperature T ,
mass density S , the Cauchy stress tensor (six components), specific internal energy u ,
&
the heat flux vector q (three components), entropy I , and pressure p , making a total of
17 unknowns.
For the problem to be well-posed 12 equations must be added to the system, as discussed
in Chapter 6, these equations are the so-called constitutive equations:
The constitutive equations
for stress:
for heat conduction:
for entropy:
The equations of state:
p1 M * Tr (D)1 2N *D
V ij
pE ij M * E ij D kk 2N *D ij
&
K ij T , j
; q K x& T (Fouriers law)
qi
I I (S , T )
p p (S , T )
u u (S , T )
(12.63)
(12.64)
(12.65)
(12.66)
where K is the thermal conductivity tensor, (see Chapter 10). So, the problem results in a
system of 17 equations with 17 unknowns.
For fluids in which is independent of temperature, the pressure can be expressed in term of
mass density, p p (S ) and the internal energy u u (S ) . So, the mechanical problem can
be represented by the following equations:
The fundamental equations for barotropic Newtonian fluids
The mass continuity equation
&
DS
S ( x& v ) 0
Dt
(12.67)
648
&
&
x& Sb Uv
(12.68)
p1 O* Tr (D)1 2N *D
(12.69)
p (S )
(12.70)
pE ij , j M * E ij D kk
In addition, by considering 2D ij
,j
2N *D ij , j
M *E ij D kk , j 2N *D ij , j
vi , j v j ,i and 2D kk
2D ij , j
vi , jj v j ,ij
D kk , j
v k ,kj
v k ,k v k , k
(12.71)
2v k ,k , we obtain:
vi , jj v j , ji
(12.72)
M *E ij D kk , j 2N *D ij , j
M *E ij v k ,kj N * vi , jj v j , ji M * v k ,ki N * vi , jj v j , ji
M * v j , ji N * vi , jj v j , ji M * N * v j , ji N * vi , jj
(12.73)
Sb i p ,i (M * N * )v j , ji N * vi , jj
Navier-Stokes-Duhem
&
&
& equations of motion
Sb x& p (M * N * ) x& ( x& v ) N * x& 2 v
(12.74)
which are the Navier-Stokes-Duhem equations of motion. The terms on the right of the
&
equation in (12.74) represents force terms, Sb represents force per unit volume and
( x& p ) is the pressure gradient and represents force per unit volume brought about by
thermodynamic pressure. Finally, the remaining terms represent the viscous force per unit
volume:
&
f vis
&
2&
(M * N * ) x& ( x& v ) N * x& v
(12.75)
12 INTRODUCTION TO FLUIDS
649
&
DS
DS
S ( x& v ) 0 ;
Sv i , i 0
Dt
Dt
*
*
Svi Sb i p ,i (M N )v j , ji N * v i , jj
&
&
&
&
Sv Sb x& p (M * N * ) x& ( x& v ) N * x& 2 v
(12.76)
(12.77)
&
*
S u pTr (D) N
>Tr(D)@2 2N *D dev : D dev
x& q Sr
2WD
(12.78)
The mass continuity equation, the Navier-Stokes-Duhem equations of motion and the
&
energy equation give us five of these in total. The unknowns are: velocity v (three
components), temperature T , mass density S , specific internal energy u , the heat flux
&
vector q (three components), entropy I , and pressure p , which makes a total of 11
unknowns.
Then, for the problem to be well-posed, six equations must be added to the system,
namely:
The constitutive equations
q i K ij T , j
(Fouriers law)
&
q K x& T
I I (S , T )
p p (S , T )
u u (S , T )
(12.79)
(12.80)
(12.81)
DS
0
S S0
Dt
Svi Sb i p ,i N * vi , jj
&
&
&
Sv Sb x& p N * x& 2 v
(12.82)
(12.83)
&
*
S u pTr (D) N
>Tr(D)@2 2N *D dev : D dev
x& q Sr
2WD
(12.84)
q i K ij T , j
(Fouriers law)
&
q K x& T
I I (S , T )
p p (S , T )
u u (S , T )
(12.85)
(12.86)
(12.87)
650
2
N * , the Navier-Stokes-Duhem equations of
3
1 *
N into the equation in (12.77) we obtain:
3
1
The Navier-Stokes equations
3
&
&
&
1
2 & of motion (Compressible fluid)
*
*
Sv Sb x& p N x& ( x& v ) N x& v
3
&
when the fluid is incompressible ( x& v v j , j 0 ) the above equation becomes:
Svi
Sb i p ,i N * v j , ji N * vi , jj
Svi
&
Sv
Sb i p ,i N * vi , jj
&
&
Sb x& p N * x& 2 v
(12.88)
(12.89)
Note that for in incompressible fluids the Navier-Stokes equations of motion (see (12.89))
and the Navier-Stokes-Duhem equations of motion (see (12.83)) coincide.
Svi
Sb i p ,i
&
Sb x& p
&
Sv
&
0 , so, the
(12.90)
&
DS
S ( x& v ) 0
Dt
&
&
Sv Sb x& p or
(12.91)
& &
& & & 1
wv ( x , t )
x& v v b x& p
(12.92)
S
wt
&
S u pTr (D) p ( x& v )
(12.93)
&
which results in a total of five equations and six unknowns, namely: v , p , u , S . Then, to
p (S , u )
(12.94)
Then, in the particular case when velocity is equal to zero, the equations of motion in
(12.98) become:
12 INTRODUCTION TO FLUIDS
&
651
Sb i
Sb x& p
p ,i
(12.95)
S dp
P( p)
(12.96)
p0
&
x& v
Svi
Sb i p , i
&
&
Sv Sb x& p
(12.97)
(12.98)
which results in four unknowns and four equations. Here, we can verify that the NavierStokes-Duhem equations of motion become the Euler equations of motion (12.98). In
addition to this, we can consider the body force field to be conservative, so, we can express
&
it by means of the potential K as follows b x& K whereby the Euler equations of
motion become:
&
&
Svi
Sv Sb x& p
S x& K x& p
Sb i p , i
SK, i p, i
(12.99)
Then, by using the material time derivative, we can express the Eulerian velocity
components as follows:
vi {
Dvi
Dt
wvi wv i
vj
wt wx j
wvi
vi , j v j
wt
(12.100)
&
&
Note, the resulting components of the operation vi , j v j are the same as those of ( x& v ) v ,
(see Chapter 1) and it was shown that the following holds:
& & 1
& & 1
& &
& & 1
( x& v ) v x& (v v ) ( x& v ) v x& (v 2 ) v x& (v 2 ) (12.101)
2
2
2
&
&
&
&
where v v , and { rot (v ) { x& v . Then, the velocity field can be expressed as
& &
( x& v ) v
follows:
&
v
&
wv & & 1 & 2
v x (v )
wt
2
(12.102)
wt
S
2
(12.103)
652
Now, if we consider that the mass density field is homogenous, i.e. it does not very with the
position vector, the following holds:
p
x&
S
1
x& ( p ) p x&
S
S
(12.104)
(12.105)
If we then consider the velocity field to be conservative, that means that the curl of the
&
&
& &
field is zero, x& v rot v 0 . Furthermore, a conservative field can be represented
by a potential, thus:
&
v
wG
x& (G) { &
wx
&
wv
wt
w wG
&
wt wx
w wG
&
wx wt
(12.106)
wG 1 2
p
x&
v K 0 i
2
S
t
w
(12.107)
Irrotacional
C (t )
(12.108)
&
0 , and if the velocity field is irrotational,:
&
0 , the equation in (12.107) becomes:
p v2
0
x& K
S 2 i
K
v2
2
constant
(12.109)
v2
2
constant
v2
p
S
Bernoullis equation
N m3
m 2 kg
Nm
kg
unit of specific energy, i.e. energy per unit mass, (see Figure 12.6).
J
kg
(12.110)
m2
, which is the
s2
12 INTRODUCTION TO FLUIDS
p v2
gh
S 2
p v2
gh
S 2
653
energy at A
energy at B
constant energy
v
2
v2
2
gh A
h
gh B
Sb i p ,i (M * N * )v j , ji N * v i , jj
&
&
&
(12.111)
Then, by taking into account the expression of velocity given in (12.102) the above
equation becomes:
Svi
&
Sb i p ,i (M * N * )v j , ji N * vi , jj
&
&
&
&
&
&
wv & & 1 & 2
2&
S
v x (v ) Sb x& p (M * N * ) x& ( x& v ) N * x& v
2
t
w
&
& 1
(M * N * ) & & & N * & 2 & &
wv & & 1 & 2
v x (v ) b x& p
x ( x v )
v 0
wt
2
S
S
S x
(12.112)
Next, we can take the curl of the above equation, the result of which is:
&
& 1
wv & & 1
(M * N * ) & & & N * & 2 &
x& v x& (v 2 ) b x& p
x ( x v )
v
S
S
S x
2
wt
>
&
&
&
&
x& x& (v 2 ) 0 , x& > x& p @ 0 , x& > x& ( x& v )@ 0 ;
&
&
& &
&
&
& &
x& >( x& v ) v @ ( x& v )( x& v ) > x& ( x& v )@ v x& v ( x& v )
;
& &
&
& &
&
&
&
x& > v @ ( x& v ) > x& @ v x& v
>
2&
x& x& v
&
&
2
2 &
x& > x& ( x& v )@ x& > x& v @ x& ;
&
0 (12.113)
654
&
wv
x&
wt
&
w
> x& v& @ w ;
wt
wt
&
If we consider that the field b is conservative, and the curl of any conservative
& &
vector field is zero, we have x& b 0 .
(12.114)
vi , i Z j vi Z j , i
(v i Z j ), j
v i , j Z j vi Z j , j
vi , i Z j
vi , j Z j
(vi Z j ), i vi Z j , i
(v i Z j ), j v i Z j , j
(12.115)
(vi Z j ), j
& &
& &
& &
( x& v ) x& > v @ ( x& ) v
&
&
& &
&
&
( x& v ) x& >v @ ( x& )v x&
(12.116)
>v @
&
&
where we have applied the definition that the divergence of the curl of a vector is zero, i.e.
&
&
x& x& ( x& v ) 0 . Then, by considering (12.116), the equation (12.114) becomes:
&
& &
& &
& & N* & 2 & &
w
0
( x& v ) ( x& ) v x& v
wt
S x
&
& &
& &
& &
& & N* & 2 & &
w
0
x& > v v @
wt
S
(12.117)
&
& &
N* & 2 & &
w
2 x& ( v ) skew
0 The equation of vorticity
wt
S x
(12.118)
>
Note that to obtain the equation in (12.118), the only simplification made was for the fluid
&
to be Newtonian, and the b -field to be conservative.
Problem 12.2: Prove that the Cauchy deviatoric stress tensor dev is equal to dev , where
V ij pE ij W ij .
Solution
If we consider that V kk 3 p W kk we can obtain:
V ijdev
V ij
V kk
G ij
3
pE ij W ij
3 p W kk
3
E ij
W ij
W kk
E ij
3
W ijdev
12 INTRODUCTION TO FLUIDS
655
Problem 12.3: Let us consider a body immersed in a Newtonian fluid. Find the total
&
traction force E acting on the closed surface S which delimits the volume V . Consider
that the bulk viscosity coefficient to be zero.
&
t (n)
t i(n ) dS
dE i
( n )
i dS
ij n j dS
ij , j dV
t i(n ) .
0 M*
2
N * (Stokes
3
condition).
Next, by considering the stress constitutive equation for Newtonian fluids, we obtain:
V ij
pE ij M * E ij D kk 2N *D ij
pE ij
2 *
N E ij D kk 2N *D ij
3
pE ij 2N * D ij kk E ij
3
Dijdev
pE ij 2N
V ij
D ijdev
Then
( pE
Ei
ij
2N *D ijdev )n j dS
( p
,i
2N *D ijdev, j )dV
where we have considered that N *,j 0 j , i.e. N * is a homogenous scalar field (homogenous
material). Then, the above equation in tensorial notation becomes:
&
E
>
&
x
p 2N * x& (D dev ) dV
(12.119)
656
Problem 12.4: Let us consider a fluid at rest which has the mass density S f . Prove
Archimedes Principle: Any body immersed in a fluid at rest experiences an upward buoyant force
equal to the weight of the volume fluid displaced by the body.
If mass density in the body is equal to S s and the body force (per unit mass) is given by
b i gE i 3 , obtain the resultant force and acceleration acting on the body.
Solution:
&
>
&
x
D dev 0 holds, and the thermodynamic pressure is equal to the hydrostatic pressure, i.e.
p p 0 whereby we have:
&
E
>
&
x
p 0 @dV
(12.120)
&
E
p0
V -volume
&
Ws
x3
x2
x1
The weight of the fluid volume displaced by the body is given by:
&
W
&
bdV
(12.121)
&
S f b
V ij , j
&
x& ( p 0 1) S f b
&
x& p 0 S f b
0i
S b i
f
( p 0 E ij ), j
p 0,i
S f b i
(12.122)
S f bi
&
bdV
&
x
p 0 dV
&
E
(12.123)
12 INTRODUCTION TO FLUIDS
&
Ws
657
&
S bdV
s
& &
E Ws
&
&
S f bdV S s bdV
(S
&
S f )bdV
Ri
(S
S f )b i dV
g (S
S f )E i 3 dV
f
s
g (S S )dV
V
thereby verifying that: if the body has a mass density lower than fluid mass density, e.g. if
& &
the body is a gas, the body rises, i.e. S f ! S s R ! 0 , and if not the body falls. Moreover,
&
&
if we consider that R m s a , where m s is the total mass of the submerged body, we can
obtain the acceleration of the body as:
s
g (S f S s )
s
f
s
f
s S
(
)
(
)
S
S
S
S
g
dV
g
dV
V
V S dV g (S f S s )
Ss
Ss
R3 V
a3
Ss
ms
ms
ms
ms
NOTE: It is interesting to note that if the medium ( f ) is such that S f 0 we have
a 3 g , i.e. the acceleration is independent of the mass. Here we have clearly seen, as did
Galileo, by means of a simple experiment, that a freely falling body was independent of the
mass. For example, on the moon where we can consider that the mass density of air is
equal to zero, two bodies with different masses in free fall, e.g. a feather and a hammer, will
have the same acceleration and will reach the moon surface at the same time.
Problem 12.5: Obtain the one-dimensional mass continuity equation for a non-viscous
incompressible fluid flow through a pipeline. Then, consider the volume V between the
two arbitrary cross sections A and B .
B
n ( B )
n ( A)
Solution:
In an incompressible medium, the mass density is independent of time
DS
{ S
Dt
&
S ( x& v )
0.
0,
658
&
x
&
v dV
k , k dV
v n dS
v n
Thus:
&
k dS
n A dS
SA
n B dS
SB
&
SA
and
SB
v A n A
&
vB
v B n B
v A n A n A dS v B n B n B dS
SA
SB
vAS A
vB S B
Problem 12.6: Determine the conditions needed for mean normal pressure
p
V kk
3
Solution:
It was deduced that:
V ijdev
2N *D ijdev
1
V kk
3
p k *D ii
V kk
3
,
p
p N *D kk
p
Thus, p
or
D ii 0
Tr (D ) 0
or
M*
2
N*
3
Bibliography
ALEXANDER, H. (1968). A constitutive relation for rubber-like material. Int. J. Eng. Sci., Vol.
6, pp. 549-563.
ANTMAN, S.S. (1995). Nonlinear Problems of Elasticity. Springer-Verlag, New York.
ARRUDA, E.M. & BOYCE, M.C. (1993). A three-dimensional constitutive model for the large
stretch behavior of rubber elastic materials. J. Mech. Phys. Solids, Vol. 41, N. 2, pp. 389412.
ASARO, R.J. & LUBARDA, V.A. (2006). Mechanics of solids and materials. Cambridge University
Press, New York, USA.
BAAR, Y. & WEICHERT (2000). Nonlinear continuum mechanics of solids: fundamental concepts and
perspectives. Springer Verlag. Berlin.
BATRA, R. C. (2006). Elements of Continuum Mechanics. John Wiley & Sons Ltd., United
Kingdom.
BAANT, Z.P. & KIM, S.-S. (1979). Plastic-Fracturing Theory for Concrete. J. Engng. mech.
Div. ASCE, 105, 407.
BAANT, Z.P. & PLANAS, J. (1997). Fracture and size effect in concrete and other quasibrittle
materials, CRC Press LLC, USA.
BIGONI, D. (2000). Bifurcation and instability of non-associative elastoplastic solids. CISM
Lecture Notes. Material Instabilities in elastic an plastic Solids, H. Petryk (IPPT, Warsaw)
Coordinator.
BIOT, M.A. (1954). Theory of stress-strain relations in anisotropic viscoelasticity and
relaxation phenomena. Jour. Appl. Phys., pp. 1385-1391.
BIOT, M.A. (1955). Variational principles in irreversible thermodynamics with application to
viscoelasticity. The Physical Reviwe 97, pp. 1463-1469.
BIOT, M.A. (1956). Thermoelasticity and irreversible thermodynamics. Jour. Appl. Phys., pp.
240-253.
BLATZ, P.J. & KO, W.L. (1962). Application of finite elasticity theory to the deformation of
rubbery material. Transactions of the Society of Rheology, Vol. VI, pp. 223-251.
BONET, J. & WOOD, R.D. (1997). Nonlinear continuum mechanics for finite element
analysis. Cambridge University Press, USA.
BUECHE, F. (1960). Molecular basis for the Mullins effect. J. Appl. Poly. Sci., 4(10), 107-114.
E.W.V. Chaves, Notes on Continuum Mechanics, Lecture Notes on Numerical
Methods in Engineering and Sciences 4, DOI 10.1007/978-94-007-5986-2,
International Center for Numerical Methods in Engineering (CIMNE), 2013
659
660
BUECHE, F. (1961). Mullins effect and rubber-filler interaction. J. Appl. Poly. Sci., 5(15), 271281.
CAROL, I. & WILLAM, K. (1997) Application of analytical solutions in elasto-plasticity to
locaization analysis of damage models. In Owen,. D.R.J., Oate, E., and Hinton, E.
(Eds.), Computational Plasticity (COMPLAS V), pp. 714-719, Barcelona. Pineridge Press.
CAROL, I.; RIZZI, E. & WILLAM, K. (1998) On the formulation of isotropic and anisotropic
damage. Computational Modelling of Concrete Structures. de borst, Biani, Mang & Meschke
(eds.). Balkema, Rotterdam, ISBN 9054109467. pp. 183-192.
CERVERA, M. & BLANCO DAZ, E. (2001). Mecnica de Estructuras Libro 1 Resistencia de
materiales. Edicions UPC, Barcelona. Eapaa.
CHABOCHE, J.L. (1979). Le concept de contrainte effective appliqu llasticit et la
viscoplasticit en presence dun endommagement anitrope. Colloque EUROMECH
115, Grenoble Edition du CNRS.
CHADWICK, P. (1976). Continuum mechanics concise theory and problems. George Allen & Unwin
Ltd.Great Britain.
CHANDRASEKHARAIAH, D.S. & DEBNATH, L. (1994). Continuum mechanics. Academic Press,
San Diego (CA, U.S.A.).
CHAVES, E.W.V.(2007). Mecnica del Medio Continuo: Conceptos Bsicos. CIMNE (Centro
Internacional de Mtodos Numricos en la Ingeniera)-Barcelona. ISBN: 978-8496736-38-2.
CHAVES, E.W.V.(2009). Mecnica del Medio Continuo: Modelos Constitutivos. CIMNE (Centro
Internacional de Mtodos Numricos en la Ingeniera)-Barcelona. ISBN: 978-8496736-68-9.
CHEN, A. & CHEN, W.F.(1975). Constitutive relations for concrete. J. Eng. Mech.-ASCE,
101:465-481.
CHEN, W.F. & HAN, D.J.(1988). Plasticity for Structural Engineers. Springer-Verlag New Yor
Inc.
CHEN, W.F. (1982). Plasticity in reinforced concrete. McGraw-Hill, Inc. USA.
CHUNG, T.J. & KIM, J.Y. (1984). Two-dimensional, combined-mode heat transfer by
conduction, convective and radiation in emitting, Absorbing, and scattering media.
ASME Trans. J. Heat Transfer 106, pp.:448-452.
CHUNG, T.J. (1996). Applied continuum mechanics, Cambridge University Press.
COLEMAN, B.D. & GURTIN, M.E. (1967). Thermodynamics with internal sate variables. J.
Chem. Phys. 47, pp. 597-613.
COLEMAN, B.D. (1964). Thermodynamics of materials with memory. Arch. Rat. Mech. Anal.
17, pp. 1-46.
CRISFIELD, M.A. (1997). Non-Linear Finite Element Analysis of Solids and Structures, volume 1,2.
John Wiley and Sons, New York, USA.
CRISTENSEN, R. M. (1982). Theory of Viscoelasticity, second edition. Dover Publications, Inc.,
New York, USA.
DESAI, C.S. & SIRIWARDANE, H.J. (1984). Constitutive laws for engineering materials with emphasis
on geological materials. Prentice-Hall, Inc.USA.
DAZ DEL VALLE, J. (1984). Mecnica de los Medios Continuos I. Servicio de publicaciones
E.T.S. de Ingenieros de Caminos, C. y P. Santander.
BIBLIOGRAPHY
661
662
LANCZOS, C. (1970). The variational principles of mechanics. Dover Publications, Inc., New
York.
LEE, E.H. (1969). Elastic-Plastic deformation at finite strains. Journal of Applied Mechanics,
Vol. 36, pp. 1-6.
LEE, E.H. (1981). Some comments on elastic-plastic analysis. Int. J. Solids Structures, Vol. 17,
pp. 859-872.
LEMAITRE, J. & CHABOCHE, J.-L. (1990). Mechanics of Solids materials. Cambridge University
Press, Cambridge.
LINDER, C. (2003). An arbritary lagrangian-Eulerian finite element formulation for dynamics and finite
strain plasticity models. Masters Thesis. University Stuttgart.
LOVE, A.E.H. (1944). A treatise on the Mathematical Theory of Elasticity. Cambridge University
Press, London.
LU, S.C.H. & PISTER, K.S. (1975). Descomposition of deformation and representation of
the free energy function for isotropic solids, Int. J. of Solids and Structures, 11, pages: 927934.
LUBARDA, V.A. & BENSON D.J. (2001). On the partitioning of the rate of deformation
gradient in phenomenological plasticity. Int. J. of Solids and Structures, 38 pages: 68056814.
LUBARDA, V.A. & KRAJCINOVIC, D. (1995). Some fundamental issues in rate theory of
damage-elastoplasticity. Int. J. of Plasticity, Vol. 11, N 7, pp. 763-797.
LUBARDA, V.A. (2004). Constitutive theories based on the multiplicative decomposition of
deformation gradient: Thermoelasticity, elastoplasticity, and biomechanics. Appl. Mech.
Rev., Vol. 57, no 2, March.
LUBLINER, J. (1990). Plasticity Theory. Macmillan Publishing Company, New York.
MALVERN, L.E. (1969). Introduction to the mechanics of a continuous medium. Prentice-Hall, Inc.
Englewood Cliffs, New Jersey.
MARSDEN, J. E. & HUGHES, T. J.R. (1983). Mathematical foundations of elasticity. Dover
Publications, Inc., New York.
MASE, G.E. (1977). Mecnica del Medio Continuo. McGraw-Hill, USA.
MAZARS, J. & LEMAITRE, J. (1984). Application of continuous damage mechanics to strain
and fracture behavior of concrete. Advances of Fracture Mechanics to Cementitious
Composites, NATO Advanced Research Workshop, 4-7 September 1984, Northwestern
University (Edited by S.P. Shah), pp. 375-388.
MAZARS, J. (1982). Mechanical damage and fracture of concrete structures. Advances in
Fracture Research (Fracture 81), Vol. 4, pp. 1499-1506. Pergamon Press, Oxford.
MENDELSON, A. (1968). Plasticity: Theory and application. New York, Robert E. Krieger
Publishing Company.
MOONEY, M. (1940). A theory of large elastic deformation. Journal of Applied Physics, Vol.
11, pp. 582-92.
MORMAN, K.N. (1986). The generalized strain measure with application to
nonhomegeneous deformation in rubber-like solids. J. Appl. Mech., Vol. 53, pp. 726728.
MORZ, Z. (1967). On the description of anisotropic work hardening. J. Mech. Phys. Solids,
15, 163.
BIBLIOGRAPHY
663
MULLINS, L. (1969). Softening of rubber by deformation. Rubber Chemistry and Tecnology, 42,
339-351.
NAGHDI, P.M. (1960). Stress-strain relations in plasticity and thermoplasticity. in E.H. Lee
& P. Symonds, eds., Plasticity. Pergamon, Oxford, pp. 121-69.
NEMAT-NASSER, S. (1982). On finite deformation elasto-plasticity. Int. J. Solids Structures,
Vol. 18, pp. 857-872.
NOWACKI, W. (1967). Problems of thermoelasticity. Progress in Thermoelasticity, VIIIth
European Mechanics Colloquium, Warszawa.
OGDEN, R.W. (1984). Non-linear elastic deformations. Dover Publications, Inc., New York.
OLIVER, J. & AGELET DE SARACBAR, C. (2000). Mecnica de medios continuos para ingenieros.
Ediciones UPC, Barcelona, Espaa.
OLIVER, J. & AGELET DE SARACBAR, C. (2000). Cuestiones y problemas de mecnica de medios
continuos. Ediciones UPC, Barcelona, Espaa.
OLIVER, J. (2002). Topics on Failure Mechanics. Monograph CIMNE N 68, International
Center for Numerical Methods in Engineering, Barcelona, Spain.
OLIVER, J.O. ; CERVERA, M. ; OLLER, S. & LUBLINER, J. (1990). Isotropic damage models
and smeared crack analysis of concrete, SCI-C Second Int. Conf. On Computer Aided Design
of Concrete Structure, Zell am See, Austria. Pg 945 957.
OLLER, S. (1988). Un modelo de dao continuo para materiales friccionales, Ph.D. thesis.
Universidad Politcnica de Catalua, Barcelona, Espaa.
OLLER, S. (2001). Fractura mecnica. Un enfoque global. CIMNE, Barcelona, Espaa.
OLLER, S.; OATE, E.; OLIVER, J. & LUBLINER, J. (1990). Finite element non-linear analysis
of concrete structures using a plastic-damage model. Engineering Fracture Mechanics Vol.
35, pp. 219-231.
OATE, E. (1992). Clculo de estructuras por el mtodo de elementos finitos anlisis esttico lineal.
Centro Internacional de Mtodos Numricos en Ingeniera. Barcelona- Espaa.
ORTIZ BERROCAL, L. (1985). Elasticidad. E.T.S. de Ingenieros Industriales. Litoprint. U.P.
Madrid.
ORTIZ, M. (1985). A Constitutive theory for inelastic behavior of concrete. Mech. Mater.,
Vol. 4, 67.
PABST, W. (2005). The linear theory of thermoelasticity from the viewpoint of rational
thermomechanics. Ceramics - Silikty, 49 (4) 242-251.
PARKER, D.F. (2003). Fields, Flows and Waves: An introduction to continuum models. SpringerVerlag London, UK.
PILKEY, W. & WUNDERLICH, W. (1992). Mechanics of structures. Variational and computational
methods. CRC Press, Inc., Florida, USA.
POWERS, J.M. (2004). On the necessity of positive semi-definite conductivity and Onsager
reciprocity in modeling heat conduction in anisotropic media. Journal of Heat Transfer,
Vol.126, pp. 670-675.
PRAGER, W. (1945). Strain hardening under combined stress. J. Appl. Phys. 16, 837-40.
PRAGER, W. (1955). The theory of plasticity: A survey of recent achievements. Proc. Inst.
Mech. Eng. 169:41.
664
BIBLIOGRAPHY
665
Index
A
acceleration
angular............................................................. 214
Eulerian ........................................................... 158
Lagrangia......................................................... 158
vector............................................................... 152
addition
vector................................................................. 18
additive decomposition
Green-Lagrange strain tensor .......................... 518
infinitesimal strain tensor ................................ 504
rate-of-deformation tensor............................... 518
adjugate of a Tensor.......................................... 42, 48
almansi strain tensor ......................181, 185, 199, 243
angle change
small deformation............................................ 232
angular momentum ............................................... 302
angular velocity vector.................................. 204, 214
angular velocity tensor.......................................... 203
anisotropic tensor.................................................... 79
anisotropy (material)............................................. 381
antisymmetric tensor......................................... 36, 38
Archimedes Principle .......................................... 656
area Element ......................................................... 215
rate................................................................... 217
associated flow...................................................... 507
auxetic materials ................................................... 399
axial vector ................................................. 37, 39, 42
axiom of Impenetrability .............................. 153, 219
B
balance of mechanical energy............................... 310
barotropic.............................................................. 638
Bauschinger effect ................................................ 366
bijective function .................................................. 149
body force ............................................................. 246
Boltzmann postulate ............................................. 303
Brazilian Test........................................................ 366
bulk modulus
adiabatic .......................................................... 564
bulk modulus ........................................................ 394
bulk viscosity coefficient ...................................... 643
C
Cauchy deformation tensor................................... 181
Cauchy heat flux................................................... 313
Cauchy stress tensor ......................................248, 250
effective (damage)........................................... 593
Cauchys first equation of motion......................... 298
Cauchys fundamental postulate........................... 248
Cauchys second law of motion............................ 303
Cauchys vorticity formula ................................... 225
Cayley-Hamilton theorem ...................................... 77
change of angle
small deformation............................................ 232
characteristic determinant....................................... 66
characteristic polynomial........................................ 67
circulation............................................................. 224
circulation preserving ........................................... 224
Clausius-Duhem inequality ...........................321, 346
Clausius-Planck inequality ................................... 321
coaxial tensors ........................................................ 81
coefficient of thermal expansion........................... 408
cofactor matrix........................................................ 49
cofactor tensor ........................................................ 42
cohesion................................................................ 368
commutative ........................................................... 81
compliance tensor ................................................. 391
component
normal ............................................................... 34
tangential ........................................................... 34
component transformation law ............................... 54
compressibility factor ........................................... 394
compressibility modulus....................................... 394
compressible hyperelasticity material................... 440
compression test ................................................... 366
triaxial ............................................................. 368
conduction ............................................................ 328
configuration
current ............................................................. 149
deformed ......................................................... 149
initial ............................................................... 149
reference.......................................................... 149
conservation Law.................................................. 291
consistency condition ........................................... 495
constitutive equations ........................................... 341
667
668
D
damage criterion ................................................... 594
damage master curve ............................................ 622
damage models ..................................................... 361
damage parameter consistency ............................. 597
damage variable............................................ 589, 598
Darcys law........................................................... 334
dashpot device ...................................................... 373
decomposition
additive.............................................................. 40
deformation
area element..................................................... 215
volume element ....................................... 215, 218
deformation gradient..................................... 163, 198
material............................................................ 165
spatial ...................................................... 166, 174
deformation of the volume element ...................... 218
deformation tensor
Cauchy............................................................. 181
left Cauchy-Green ............................177, 198, 212
Piola ........................................................ 177, 184
right Cauchy-Green ..........................177, 183, 191
density .................................................................. 285
derivative with tensors............................................ 84
derivative ................................................................ 94
descriptions
Eulerian ........................................................... 152
Lagrangian....................................................... 152
determinant of a tensor ........................................... 45
deviatoric plane..................................................... 139
deviatoric tensor
Voigt notation.................................................. 101
E
effective stress ...............................................587, 589
eigenvalue........................................................66, 127
eigenvector ..............................................................66
Einstein notation......................................................20
elastic acoustic tensor ............................................398
elastic compliance tensor...............................391, 397
elastic limit ............................................................363
elastic modulus ......................................................363
elastic potential......................................................424
elastic process................................................357, 360
elastic pseudomoduli .............................................431
elastic stiffness tensor............................................377
elasticity tensor..............................358, 377, 390, 397
isothermal.........................................................557
tangent..............................................................427
elasticity tensor components..................................381
elastoplastic model
isotropic-kinematic...........................................502
kinematic hardening .........................................501
small deformation.............................................507
ellipsoid
tensor................................................................138
energy density........................................................286
strain.................................................................357
energy equation .....................................................315
Eulerian ............................................................315
Lagrangian .......................................................317
with discontinuities ..........................................317
engineering notation ..............................................375
engineering strain ..........................................233, 243
enthalpy .........................................................322, 343
entropy...................................................................319
entropy inequality..................................................319
with discontinuity.............................................325
equation of vorticity ..............................................654
equations of motion .......................................297, 376
Eulerian ............................................................298
Lagrangian .......................................................298
with discontinuities ..........................................301
with discontinuities ..........................................302
INDEX
equations of state .................................................. 343
equilibrium equations
Eulerian ........................................................... 298
Lagrangian....................................................... 299
equivalent plastic strain ........................................ 513
Euclidean norm....................................................... 11
Euler angles ............................................................ 65
Eulerian
variables .......................................................... 153
Eulerian description .............................................. 157
Eulerian finite strain tensor................................... 181
Eulerian stretch tensor .......................................... 195
Eulerian variable
material time derivative ................................... 158
external mechanical power ................................... 308
F
Ficks law of diffusion.......................................... 335
field
conservative............................................. 115, 120
scalar ............................................................... 105
second-order tensor ......................................... 105
stationary ......................................................... 159
vector............................................................... 105
finite strain............................................................ 176
first law of thermodynamics ......................... 307, 314
first Piola-Kirchhoff stress tensor ................. 261, 262
flow plastic vector................................................. 468
flow rule
associated ........................................................ 494
perfect plasticity .............................................. 493
Prandtl-Reuss................................................... 517
fluids ..................................................................... 369
flux........................................................................ 286
energy.............................................................. 287
mass................................................................. 287
flux problem ................................................. 327, 338
forces .................................................................... 245
body................................................................. 246
density ............................................................. 246
gravitational..................................................... 246
internal............................................................. 245
surface ............................................................. 245
thermodynamic ........................................ 356, 505
viscous............................................................. 648
Fouriers equation................................................. 333
Fouriers law of heat conduction .......... 325, 328, 329
fourth-order projection tensor ................................. 54
fragile materials .................................................... 364
Frobenius norm....................................................... 79
fundamental equations of continuum mechanics .. 342
G
Gauss theorem ..................................................... 117
generalized Hookes law....................................... 377
Gibbs free energy.......................................... 322, 343
gradient ................................................................. 106
Green elasticity ..................................................... 423
Greens first identity ............................................. 123
Greens second identity ........................................ 123
Greens theorem.................................................... 121
Green-Lagrange strain tensor........................ 178, 243
669
Green-Naghdi Rate ............................................... 282
H
Haigh-Westergaard stress space ........................... 470
hardening plasticity............................................... 490
hardening/softening modulus
continuum........................................................ 597
heat capacity ......................................................... 325
heat conduction inequality.............................321, 547
heat flux equation ..........................................325, 332
heat flux.........................................................313, 329
heat source............................................................ 314
Helmholtz free energy ...................................322, 343
isotropic damage model................................... 591
hexagonal symmetry............................................. 386
homogeneous deformation.................................... 191
homogeneous........................................................ 105
continuum........................................................ 146
Hookes law...........................................372, 377, 392
plane strain ...................................................... 414
plane stress ...................................................... 412
Voigt notation...........................................378, 379
hydrostatic axis
140
hydrostatic pressure .......................................368, 636
hyperelasticity................................................361, 423
I
identity tensor ......................................................... 44
incompressibility restriction ................................. 450
incompressible...............................................220, 295
index
dummy .............................................................. 21
free .................................................................... 21
inelastic behavior.................................................. 361
infinitesimal strain tensor ..............................230, 375
infinitesimal strain theory ..................................... 229
initial boundary value problem ............................. 339
linear elasticity .........................................357, 376
initial configuration .............................................. 145
initial strain........................................................... 412
instantaneous elastic moduli ................................. 431
integration by parts ............................................... 117
internal energy ...................................................... 313
internal force......................................................... 245
internal friction ..................................................... 368
internal variables .................................................. 506
invariant.......................................................43, 68, 86
irreversible process ............................................... 362
irrotational .....................................................173, 224
isentropic process ..........................................319, 552
isochoric motion ....................................220, 225, 295
isochoric transformation ....................................... 440
isothermal Lam constants ................................... 562
isothermal processes ............................................. 552
isotropic material ...........................................359, 390
isotropic tensor ....................................................... 80
isotropic tensor function ......................................... 92
isotropic-kinematic hardening plasticity............... 491
isotropy................................................................. 381
J
Jacobian determinant .....................153, 174, 175, 220
670
K
Kelvin-Stokes theorem ........................................ 121
kinematic equations .............................................. 377
kinematic hardening plasticity .............................. 490
kinematic tensors .......................................... 183, 197
rate................................................................... 208
objectivity........................................................ 273
kinetic energy ....................................................... 307
rigid body motion ............................................ 312
Knudsen number................................................... 635
Kronecker delta............................................22, 44, 80
Kuhn strain ........................................................... 244
Kuhn-Tucker conditions ....................................... 494
L
Lagrangian
variable............................................................ 153
Lagrangian description ......................................... 156
Lagrangian stretch tensor...................................... 195
Lam constants ..............................359, 363, 390, 397
isentropic ......................................................... 553
isothermal ........................................................ 553
Laplaces equation........................................ 333, 335
Laplacian operator ................................................ 112
latent heat tensor....................................554, 557, 572
left Cauchy-Green deformation tenso ................... 198
left stretch tensor .................................................. 195
Levi-Civita pseudo-tensor................................. 45, 80
Lie derivative........................................................ 528
linear elasticity...................................................... 361
linear elasticity theory........................................... 375
linear momentum.................................................. 297
rigid body motion ............................................ 304
linearly dependent................................................... 13
M
Macaulay bracket.................................................. 603
mass continuity equation .............................. 224, 291
Eulerian ........................................................... 292
incompressible medium................................... 295
Lagrangian....................................................... 294
with discontinuities.......................................... 295
mass density...................................146, 150, 285, 286
material curve ........................................169, 194, 220
material frame indifference................................... 270
material point........................................................ 146
material surface..................................................... 220
material time derivative ........................................ 156
material volume .................................................... 221
maximum tangential component
129
mean stress............................................................ 636
mechanical power ................................................. 309
mechanical properties ........................................... 398
mixed product......................................................... 12
Mohrs circle ................................................ 134, 137
moments of inertia ................................................ 305
monoclinic symmetry ........................................... 383
motion with deformation ...................................... 147
Mullins effect........................................................ 621
multiplication
scalar ..................................................................29
multiplicative decomposition
deformation gradient ........................................518
volumetric and isochoric ..................................225
N
nabla symbol..........................................................106
Nansons formula ..........................................216, 262
Navier-Lam equations..........................................410
Navier-Poisson law................................................643
negative definite (tensor) .........................................52
Neumann boundary condition................................333
Newtons law of cooling........................................330
Newtonian fluids ...................................................371
non-linear elastic ...................................................361
nonlinear elasticity.................................................423
non-Newtonian fluids ............................................371
normal component .................................................127
normal octahedral vector .......................................139
normal vector.........................................................125
normalization condition.................................427, 436
norms of tensors ......................................................79
notation
indicial..........................................................20, 34
symbolic .............................................................34
tensorial..............................................................34
O
objective rates........................................................278
objectivity of tensors .............................................270
octahedral plane.............................................139, 140
octahedral vector
tangent..............................................................139
Oldroyd rate...........................................................279
operator
Laplacian ..........................................................112
orthogonal matrix ..................................................148
orthogonal tensor.............................................51, 148
improper .............................................................51
proper .................................................................51
orthogonal transformation .......................................51
orthogonality ...........................................................12
orthonormal basis ..............................................16, 26
orthotropic material ...............................................404
orthotropic symmetry ............................................384
ortogonalidad...........................................................51
P
parallel axis theorem..............................................306
particle...................................................................146
path line .................................................................146
permutation symbol .........................................23, 113
permutation tensor ...................................................45
persistency
condition...........................................................495
Piola deformation tensor................................177, 184
plane strain ....................................................236, 410
plane stress ............................................................410
plastic flow rule .....................................................506
plastic flow tensor..................................................468
plastic multiplier....................................................493
INDEX
plasticity
perfect.............................................................. 491
plasticity models ................................................... 361
elastic-perfectly ............................................... 489
Poissons equation ................................................ 332
Poissons ratio....................................................... 394
polar decomposition........................................ 82, 195
rates ................................................................. 203
rotation tensor.................................................. 198
polar rate............................................................... 282
pore pressure......................................................... 369
position vector ...................................................... 147
positive definite tensor.................................... 52, 126
power extended..................................................... 310
principal invariants ........................................... 67, 68
derivative ........................................................... 94
deviatoric stress tensor..................................... 259
principal space ................................................ 67, 128
principle
conservation of angular momentum ................ 285
conservation of energy............................. 285, 307
conservation of linear momentum ................... 285
conservation of mass ....................................... 285
determinism ..................................................... 343
dissipation........................................................ 343
equipresence .................................................... 343
irreversibility ................................................... 285
limited memory ............................................... 344
local action ...................................................... 343
objectivity........................................................ 343
of action and reation ........................................ 249
of objectivity.................................................... 270
Saint-Venant.................................................... 406
superposition ................................................... 407
product
tensor ................................................................. 28
double scalar...................................................... 30
products of inertia ................................................. 305
projection tensor ................................................... 441
proportionality limit.............................................. 363
pseudo-invariants of anisotropy............................ 463
pseudo-tensor
Levi-Civita ........................................................ 24
R
radiant heat constant ............................................. 314
radiation................................................................ 328
ramp function........................................................ 603
rate
convective........................................................ 279
Cotter-Rivlin.................................................... 280
Green-McInnis................................................. 282
Green-Naghdi .................................................. 282
Jaumann-Zaremba ........................................... 280
Oldroyd.................................................... 279, 281
polar................................................................. 282
Truesdell.......................................................... 283
rate independent.................................................... 467
rate of change
convective........................................................ 157
local ................................................................. 157
rate of the material rotation tensor ........................ 203
rate-of-deformation tensor ............................ 172, 204
recoverable power................................................. 646
reflection tensor ...................................................... 51
671
relaxation .............................................................. 372
reversible process ..........................................323, 427
Reynolds number.................................................. 639
Reynolds transport theorem................................. 287
with discontinuities ......................................... 290
rheological models................................................ 372
isotropic hardening elastoplastic ..................... 496
perfect elastoplastic ......................................... 492
right Cauchy-Green deformation tensor ........177, 183
right stretch tensor ................................................ 195
rigid body motion .. 147, 148, 173, 182, 192, 214, 327
rotation tensor......................................................... 51
infinitesimal..................................................... 230
polar decomposition ........................................ 198
rate................................................................... 203
rotor ...................................................................... 113
rupture strength point............................................ 363
S
scalar......................................................................... 9
scalar multiplication ............................................... 18
scalar product.........................................11, 18, 22, 30
Voigt notation.................................................... 98
scalar product trace................................................. 43
scalar triple product .......................................... 12, 18
scalar-valued tensor ................................................ 91
second law of Thermodynamics ........................... 319
second Piola-Kirchhoff stress tensor .................... 262
second-order tensor
projection......................................................... 125
second-order-valued tensor..................................... 91
shear modulus....................................................... 394
sign function ......................................................... 494
simple thermoelastic materials.............................. 344
singular
tensor................................................................. 46
skew tensor ............................................................. 36
small deformation theory...................................... 229
spatial velocity gradient........................................ 213
specific heat ...................................................560, 572
constant stress.................................................. 554
constant volume............................................... 554
specific internal energy......................................... 343
spectral representation
Voigt notation.................................................. 100
spectral representation .................................73, 74, 81
spherical axis ........................................................ 140
spherical part .......................................................... 54
spherical tensor..........................................68, 86, 138
spin tensor............................................................. 172
infinitesimal..................................................... 230
spring device......................................................... 373
Stefan-Boltzmann law .......................................... 330
Steiners theorem.................................................. 306
stiffness modulus
damage secant ................................................. 589
stiffness tensor
elastic .............................................................. 377
elastic-damage secant ...................................... 593
Stokes condition .................................................. 644
Stokesian fluids .................................................... 371
strain
engineering...................................................... 233
equivalent ........................................................ 594
strain energy density......................357, 400, 402, 424
672
T
tangent stiffness modulus
elastoplastic ............................................. 499, 501
tangent stiffness pseudo-tensor
elastic ...................................................... 431, 432
tangent stiffness tensors........................................ 433
INDEX
transformation law
elasticity tensor................................................ 381
transformation law ............................................ 54, 58
transformation matrix ..........................56, 62, 99, 380
transport equations........................................ 223, 338
transpose ................................................................. 34
transversely isotropic material .............................. 405
transversely isotropic symmetry ........................... 388
triclinic materials .................................................. 382
triple scalar product ................................................ 26
true strain ...................................................... 241, 243
true stress tensor ................................................... 250
Truesdell stress rate .............................................. 283
U
ultimate strength point .......................................... 363
undeformed configuration..................................... 145
uniform ................................................................. 105
unit extension................................................ 164, 188
small deformation............................................ 233
small deformation............................................ 231
unit tensor
fourth-order ................................................. 44, 75
second-order ...................................................... 44
V
vector ........................................................................ 9
angular-velocity............................................... 204
axial ................................................................... 37
norm of .............................................................. 18
octahedral
139
product......................................12, 18, 25, 32, 113
projection........................................................... 11
triple product ............................................... 13, 19
unit............................................................... 11, 18
vorticity ........................................................... 172
zero.................................................................... 11
zero.................................................................... 18
velocity
angular............................................................. 214
Eulerian ........................................................... 158
Lagrangian....................................................... 158
vector............................................................... 152
velocity gradient
spatial ...................................................... 172, 213
virgin curve........................................................... 622
viscoelastic materials ............................................ 371
viscosity................................................................ 370
viscous forces ....................................................... 648
Voigt notation ................................................... 96, 97
transfromation law............................................. 99
unit tensors ........................................................ 97
volume element
rate................................................................... 219
volume element deformation ................................ 218
volume ratio
small deformation............................................ 235
vorticity vector...................................... 173, 224, 640
Y
yield condition
Huber-von Mises ............................................. 512
673
Mises-Huber.................................................... 546
yield criterion........................................................ 468
Alternative Drucker-Prager ............................. 484
yield curve ............................................................ 470
yield point............................................................. 363
yield surface
Drucker-Prager ................................................ 484
isotropic material............................................. 469
Mohr-Coulomb................................................ 480
Rankine ........................................................... 488
von Mises ........................................................ 474
Youngs modulus.................................................. 363