You are on page 1of 364

CONTINUUM

MECHANICS
FOR ENGINEERS

08/07/2017 Dr. Luis Mosquera L. 1


Contents
1. Continuum Theory
• The Continuum Concept
• Continuum Mechanics
2. Essential Mathematics
• Scalars, Vectors, and Cartesian Tensors
• Tensor Algebra
• Summation Convention
• Indicial Notation
• Matrices and Determinants
• Transformations of Cartesian Tensors
• Principal Values and Principal Directions of Symmetric
• Second-Order Tensors
• Tensor Fields, Tensor Calculus
• Integral Theorems of Gauss and Stokes

08/07/2017 Dr. Luis Mosquera L. 2


3. Stress Principles
• Body and Surface Forces, Mass Density
• Cauchy Stress Principle
• The Stress Tensor
• Force and Moment Equilibrium,
• Stress Tensor Symmetry
• Stress Transformation Laws
• Principal Stresses, Principal Stress Directions
• Maximum and Minimum Stress Values
• Mohr’s Circles for Stress
• Plane Stress
• Deviator and Spherical Stress States
• Octahedral Shear Stress

08/07/2017 Dr. Luis Mosquera L. 3


4. Kinematics of Deformation and Motion
• Particles, Configurations, Deformation, and Motion
• Material and Spatial Coordinates
• Lagrangian and Eulerian Descriptions
• The Displacement Field
• The Material Derivative
• Deformation Gradients, Finite Strain Tensors
• Infinitesimal Deformation Theory
• Stretch Ratios
• Rotation Tensor, Stretch Tensors
• Velocity Gradient, Rate of Deformation, Vorticity
• Material Derivative of Line Elements, Areas, Volumes

08/07/2017 Dr. Luis Mosquera L. 4


5. Strain Tensors

• The deformation gradient tensor


• The Lagrangian finite strain tensor
• The Eulerian finite strain tensor
• Displacement gradients
• Infinitesimal deformation theory

08/07/2017 Dr. Luis Mosquera L. 5


6. Fundamental Laws and Equations
• Balance Laws, Field Equations, Constitutive Equations
• Material Derivatives of Line, Surface, and Volume Integrals
• Conservation of Mass, Continuity Equation
• Linear Momentum Principle, Equations of Motion
• The Piola-Kirchhoff Stress Tensors,
• Lagrangian Equations of Motion
• Moment of Momentum (Angular Momentum) Principle

08/07/2017 Dr. Luis Mosquera L. 6


7. Law of Conservation of Energy, The Energy Equation
• Entropy and the Clausius-Duhem Equation
• Restrictions on Elastic Materials by the Second
• Law of Thermodynamics
• Invariance
• Restrictions on Constitutive Equations
• from Invariance
• Constitutive Equations

08/07/2017 Dr. Luis Mosquera L. 7


8. Linear Elasticity
• Elasticity, Hooke’s Law, Strain Energy
• Hooke’s Law for Isotropic Media, Elastic Constants
• Elastic Symmetry; Hooke’s Law for Anisotropic Media
• Isotropic Elastostatics and Elastodynamics,
• Superposition Principle
• Plane Elasticity
• Linear Thermoelasticity
• Airy Stress Function
• Torsion
• Three-Dimensional Elasticity

08/07/2017 Dr. Luis Mosquera L. 8


9. Classical Fluids

• Viscous Stress Tensor, Stokesian, and Newtonian Fluids


• Basic Equations of Viscous Flow, Navier-Stokes Equations
• Specialized Fluids
• Steady Flow, Irrotational Flow, Potential Flow
• The Bernoulli Equation, Kelvin’s Theorem

08/07/2017 Dr. Luis Mosquera L. 9


BIBLIOGRAFIA
[1] G. Thomas Mase, Ronald Smelser. Continuum Mechanics for Engineers. CRC
Press 2010
[2] Xavier Oliver Olivella, Carlos Agelet de Saracibar Bosch . Mecánica de
Medios Continuos para Ingenieros. Ed. Alfa Omega. 2002.
[3] I. S. Sokolnikoff. Mathematical Theory of Elasticity. Ed. McGraw-Hill.1956.

08/07/2017 Dr. Luis Mosquera L. 10


Continuum Theory

A body is really a collection of discrete atoms stacked on


one another in a particular repetitive lattice.
Thus, matter is not continuous.

08/07/2017 Dr. Luis Mosquera L. 11


Material Body

CRISTALINOS

AMORFO

08/07/2017 Dr. Luis Mosquera L. 12


The continuum model for material bodies is important for two very good
reasons:

- On the scale by which we consider bodies, the characteristic dimensions


are extremely large compared to molecular distances.

- Our knowledge of the mechanical behavior of materials is based almost


entirely upon experimental data gathered by tests on relatively large
specimens.

08/07/2017 Dr. Luis Mosquera L. 13


Continuum Mechanics
The analysis of the mechanical behavior of materials modeled on the
continuum assumption is what we know as continuum mechanics.

- The derivation of fundamental equations which are valid for all


continuous media. These equations are based upon universal laws of
physics such as the conservation of mass, the principles of energy and
momentum, etc.

- The development of so-called constitutive equations characterizing the


behavior of specific idealized materials, the perfectly elastic solid and the
viscous fluid being the best known examples.

08/07/2017 Dr. Luis Mosquera L. 14


Essential Mathematics

A tensor, or linear transformation, assigns any vector v another vector Tv


such that:
T(u+w) = Tu + Tw
T(αv) = α (Tv)
Para todo v y w
Furthermore:

(T+S)v = Tv + Sv
(αT)v = α (Tv)

Because of these properties, tensors constitute a vector space

08/07/2017 Dr. Luis Mosquera L. 15


Scalars, Vectors, and Cartesian Tensors

As it happens, a considerable variety of physical and


geometrical quantities have important roles in continuum
mechanics, and fortunately, each of these may be
represented by some form of tensor.

08/07/2017 Dr. Luis Mosquera L. 16


For example, such quantities as density and Temperature may be
specified completely by giving their magnitude, i.e., by stating a
numerical value. These quantities are represented mathematically by
scalars, which are referred to as zeroth-order tensors.

08/07/2017 Dr. Luis Mosquera L. 17


Several physical quantities of mechanics such as force require not only an
assignment of magnitude, but also a specification of direction for their
complete characterization. Quantities possessing such directional properties
are represented by vectors, which are first-order tensors.

08/07/2017 Dr. Luis Mosquera L. 18


A significant number of physical quantities having important status in
continuum mechanics require mathematical entities of higher order than
vectors for their representation in the hierarchy of tensors. As we shall
see, among the best known of these are the stress tensor and the strain
tensors. These particular tensors are second-order tensors, and are said
to have a rank of two.

08/07/2017 Dr. Luis Mosquera L. 19


Tensors, like vectors, are independent of any coordinate system,
but just as with vectors, when we wish to specify a tensor by its
components we are obliged to refer to a suitable set of reference
axes.

08/07/2017 Dr. Luis Mosquera L. 20


Tensor Algebra in Symbolic Notation —Summation
Convention
The three-dimensional physical space of everyday life is the space in
which many of the events of continuum mechanics occur.
Mathematically, this space is known as a Euclidean three-space, and its
geometry can be referenced to a system of Cartesian coordinate axes.

08/07/2017 Dr. Luis Mosquera L. 21


Unit vectors in the coordinate directions
x1,x2, and x3.

Rectangular components of the vector v

08/07/2017 Dr. Luis Mosquera L. 22


Because a scalar has only a single component, it will have
the same value in every system of axes, but the
components of vectors and tensors will have different
component values, in general, for each set of axes.

08/07/2017 Dr. Luis Mosquera L. 23


identify with these axes the triad of unit base vectors 𝑒1 ,
𝑒2 , 𝑒3 , due to the mutual perpendicularity of these base
vectors, they form an orthogonal basis; furthermore,
because they are unit vectors, the basis is said to be
orthonormal.

𝑽 = 𝑣1 𝑒1 + 𝑣2 𝑒2 + 𝑣3 𝑒3 = 𝑣𝑖 𝑒𝑖
𝑖=1

08/07/2017 Dr. Luis Mosquera L. 24


the summation convention. Stated briefly, we agree that
whenever a subscript appears exactly twice in a given term,
that subscript will take on the values 1, 2, 3 successively,
and the resulting terms summed.

𝑉 = 𝑣𝑖 𝑒𝑖

08/07/2017 Dr. Luis Mosquera L. 25


Example: expand the following expression according to the
summation convention:

𝑢𝑖 𝑣𝑖 𝑤𝑗 𝑒𝑗

08/07/2017 Dr. Luis Mosquera L. 26


Solution:

Summing first on i, and then on j,

𝑢𝑖 𝑣𝑖 𝑤𝑗 𝑒𝑗 = 𝑢1 𝑣1 + 𝑢2 𝑣2 + 𝑢3 𝑣3 𝑤1 𝑒1 + 𝑤2 𝑒2 + 𝑤3 𝑒3

08/07/2017 Dr. Luis Mosquera L. 27


Example: expand the following expression according to the
summation convention:

Solution:
Summing first on i, and then on j,

= 𝑻𝟏𝟏 𝒗𝟏 + 𝑻𝟐𝟏 𝒗𝟐 + 𝑻𝟑𝟏 𝒗𝟑 𝒆𝟏 + 𝑻𝟏𝟐 𝒗𝟏 + 𝑻𝟐𝟐 𝒗𝟐 + 𝑻𝟑𝟐 𝒗𝟑 𝒆𝟐


+ 𝑻𝟏𝟑 𝒗𝟏 + 𝑻𝟐𝟑 𝒗𝟐 + 𝑻𝟑𝟑 𝒗𝟑 𝒆𝟑

08/07/2017 Dr. Luis Mosquera L. 28


Kronecker Delta

1 𝑖=𝑗
𝛿𝑖𝑗 =
0 𝑖≠ 𝑗

Note that:
𝛿𝑖𝑖 = 𝛿11 + 𝛿22 + 𝛿33 = 3

𝛿𝑖𝑗 𝑒𝑗 = 𝑒𝑖

08/07/2017 Dr. Luis Mosquera L. 29


Permutation Symbol

1 𝑖𝑓 𝑖𝑗𝑘 𝑎𝑝𝑝𝑒𝑎𝑟 𝑎𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒 12312


𝜀𝑖𝑗𝑘 = −1 𝑖𝑓 𝑖𝑗𝑘 𝑎𝑝𝑝𝑒𝑎𝑟 𝑎𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒 32132
0 𝑖𝑓 𝑖𝑗𝑘 𝑎𝑝𝑝𝑒𝑎𝑟 𝑖𝑛 𝑎𝑛𝑦 𝑜𝑡ℎ𝑒𝑟 𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒

The cross product:


𝑒𝑖 × 𝑒𝑗 = 𝜀𝑖𝑗𝑘 𝑒𝑘

ε-δ identity: 𝜀𝑚𝑖𝑞 𝜀𝑗𝑘𝑞 = 𝛿𝑚𝑗 𝛿𝑖𝑘 − 𝛿𝑚𝑘 𝛿𝑖𝑗

08/07/2017 Dr. Luis Mosquera L. 30


Several useful definitions from vector/tensor algebra.

1. Addition of vectors:
𝑤 =𝑢+𝑣
or
𝑤𝑖 𝑒𝑖 = (𝑢𝑖 + 𝑣𝑖 )𝑒𝑖

2. Multiplication:
(a) of a vector by a scalar:

𝜆𝑣 = 𝜆𝑣𝑖 𝑒𝑖
(b) dot (scalar) product of two vectors:

𝑢. 𝑣 = 𝑢𝑖 𝑣𝑖

08/07/2017 Dr. Luis Mosquera L. 31


(c) The vector cross product:

𝑢 × 𝑣 = 𝑢𝑖 𝑒𝑖 × 𝑣𝑗 𝑒𝑗 = 𝜀𝑖𝑗𝑘 𝑢𝑖 𝑣𝑗 𝑒𝑘

(d) The triple scalar product:

𝑢. 𝑣 × 𝑤 = 𝜀𝑗𝑘 𝑞 𝑢𝑖 𝑣𝑗 𝑤𝑘 𝛿𝑖𝑞 = 𝜀𝑖𝑗 𝑘 𝑢𝑖 𝑣𝑗 𝑤𝑘

(e) The triple cross product:

𝑢 × (𝑣 × 𝑤) = (𝛿𝑚𝑗 𝛿𝑖𝑘 − 𝛿𝑚𝑘 𝛿𝑖𝑗 ) 𝑢𝑖 𝑣𝑗 𝑤𝑘 𝑒𝑚

08/07/2017 Dr. Luis Mosquera L. 32


(f) Dyad: Two vectors can be multiplied together to yield a tensor. The
tensor product of two vectors creates a dyad.

𝑢𝑣 = 𝑢𝑖 𝑒𝑖 𝑣𝑗 𝑒𝑗 = 𝑢𝑖 𝑣𝑗 𝑒𝑖 𝑒𝑗

𝑢𝑖 𝑣𝑗 𝑒𝑖 𝑒𝑗 = 𝑢1 𝑣1 𝑒1 𝑒1 + 𝑢1 𝑣2 𝑒1 𝑒2 + 𝑢1 𝑣3 𝑒1 𝑒3 + 𝑢2 𝑣1 𝑒2 𝑒1 + 𝑢2 𝑣2 𝑒2 𝑒2 + 𝑢2 𝑣3 𝑒2 𝑒3
+ 𝑢3 𝑣1 𝑒3 𝑒1 + 𝑢3 𝑣2 𝑒3 𝑒2 + 𝑢3 𝑣3 𝑒3 𝑒3

Nonion form of the dyad

08/07/2017 Dr. Luis Mosquera L. 33


A sum of dyads such as: 𝒖𝟏 𝒗𝟏 + 𝒖𝟐 𝒗𝟐 + ⋯ + 𝒖𝒏 𝒗𝒏

Is called a dyadic.

Alternative notation used for the dyad product is:

𝑎1 𝑢1
𝑎0𝑏 𝑢 = 𝑎 𝑏. 𝑢 = 𝑎2 𝑏1 𝑏2 𝑏3 𝑢2
𝑎3 𝑢3

(g) Dyads can be multiplied by each other to yield another dyad:

𝑢𝑣 . 𝑤𝑠 = 𝑢𝑖 𝑒𝑖 (𝑣𝑗 𝑒𝑗 . 𝑤𝑘 𝑒𝑘 )𝑆𝑞 𝑒𝑞 = 𝑢𝑖 𝑣𝑗 𝑤𝑗 𝑆𝑞 𝑒𝑖 𝑒𝑞

08/07/2017 Dr. Luis Mosquera L. 34


(h) Vectors can be multiplied by a tensor to give a vector:

𝑣. 𝑇 = 𝑣𝑖 𝑒𝑖 . 𝑡𝑗𝑘 𝑒𝑗 𝑒𝑘 = 𝑣𝑖 𝑡𝑗𝑘 𝛿𝑖𝑗 𝑒𝑘 = 𝑣𝑖 𝑡𝑖𝑘 𝑒𝑘

𝑇. 𝑣 = 𝑡𝑖𝑗 𝑒𝑖 𝑒𝑗 . 𝑣𝑘 𝑒𝑘 = 𝑡𝑖𝑗 𝑒𝑖 𝛿𝑗 𝑘 𝑣𝑘 = 𝑡𝑖𝑗 𝑣𝑗 𝑒𝑖

(i) Two tensors can be multiplied resulting in a tensor:

𝑇. 𝑆 = 𝑡𝑖𝑗 𝑒𝑖 𝑒𝑗 . 𝑠𝑝𝑞 𝑒𝑝 𝑒𝑞 = 𝑡𝑖𝑗 𝑠𝑝𝑞 𝛿𝑗𝑝 𝑒𝑖 𝑒𝑞 = 𝑡𝑖𝑗 𝑠𝑗𝑞 𝑒𝑖 𝑒𝑞

08/07/2017 Dr. Luis Mosquera L. 35


One of the most important advantages of the indicial notation
is the compactness it provides in expressing equations in three
dimensions:

𝑡𝑖𝑗 = λ𝛿𝑖𝑗 𝐸𝑘𝑘 + 2𝜇𝐸𝑖𝑗

9 equations, 4 terms of each

08/07/2017 Dr. Luis Mosquera L. 36


Problems
1. Let v = a x b
Using indicial notation, show that,

a) 𝑣. 𝑣 = 𝑎2 𝑏2 𝑠𝑖𝑛2 𝜃
b) 𝑎 × 𝑏. 𝑎 = 0
c) 𝑎 × 𝑏. 𝑏 = 0

2. If 𝐴𝑖𝑗 = 𝛿𝑖𝑗 𝐵𝑘𝑘 + 3 𝐵𝑖𝑗, Determine Bkk and using that solve for Bij in terms of Aij

and its first invariant, Aii

3. Use indicial notation to show that:

𝐴𝑚𝑖 𝜀𝑚𝑗𝑘 + 𝐴𝑚𝑗 𝜀𝑖𝑚𝑘 +𝐴𝑚𝑘 𝜀𝑖𝑗 𝑚 = 𝐴𝑚𝑚 𝜀𝑖𝑗𝑘

Where A is any tensor


08/07/2017 Dr. Luis Mosquera L. 37
4. By summing on the repeated subscripts determine the simplest form of:

(a) 𝜀3𝑗𝑘 𝑎𝑗 𝑎𝑘
(b) 𝜀𝑖𝑗𝑘 𝛿𝑗𝑘
(c) 𝜀1𝑗𝑘 𝑎2 𝑇𝑘𝑗

5. Consider the tensor 𝐵𝑖𝑘 = 𝜀𝑖𝑗𝑘 𝑣𝑗 , show that Bik is skew-symmetric

08/07/2017 Dr. Luis Mosquera L. 38


Indicial notation

λ= scalar (zeroth order tensor)


Vi = vector (first order tensor)
UiVj = dyad (second order tensor)
Tij = dyadic (second order tensor)
Qijk = triadic (third order tensor)
Cijkm = tetradic (forth order tensor)

For tensors defined in a three-dimensional space, the free indices take on the values
1,2,3 successively, and we say that these indices have a range of three. If N is the
number of free indices in a tensor, that tensor has 3𝑁components in three space.

08/07/2017 Dr. Luis Mosquera L. 39


In the indicial notation exactly two types of subscripts appear:

1. “Free” indices, wich are represented by letters that occur only once in a
given term,
2. “summed”, or “dummy” indices which are represented by letters that
appear only twice in a given term.

No letter subscript may appear more than


twice in any given term.

Mathematical operations among tensors are readily carried out


using the indicial notation:
𝑢𝑖 + 𝑣𝑖 − 𝑤𝑖 = 𝑠𝑖
𝑡𝑖𝑗 − 𝑣𝑖𝑗 + 𝑠𝑖𝑗 = 𝑞𝑖𝑗

08/07/2017 Dr. Luis Mosquera L. 40


Multiplication of two tensors to produce an outer tensor product is
accomplished by simply setting down the tensor symbols side by side with no
dummy indices appearing in the expression.

Example: the outer product of the vector Vi and tensor tjk is the third-order
tensor Vitjk

Contraction is the process of identifying (that is, setting equal to one another)
any two indices of a tensor term.

An inner tensor product is formed from an outer tensor product by one or


more contractions involving indices from separate tensors in the outer
product. The rank of a given tensor is reduced by two for each contraction.

08/07/2017 Dr. Luis Mosquera L. 41


Forms for inner and outer products

Outer Products Contraction (s) Inner Products


𝑢𝑖 𝑣𝑗 𝑖=𝑗 𝑢𝑖 𝑣𝑖 Vector dot product

𝜀𝑖𝑗𝑘 𝑢𝑞 𝑣𝑚 𝑗 = 𝑞, 𝑘 = 𝑚 𝜀𝑖𝑗𝑘 𝑢𝑗 𝑣𝑘 Vector cross product

𝜀𝑖𝑗𝑘 𝑢𝑞 𝑣𝑚 𝑤𝑛 𝑖 = 𝑞, 𝑗 = 𝑚, 𝑘 = 𝑛 𝜀𝑖𝑗𝑘 𝑢𝑖 𝑣𝑗 𝑤𝑘 Box product

08/07/2017 Dr. Luis Mosquera L. 42


Tensor symmetric
𝑠𝑖𝑗 = 𝑠𝑗𝑖

𝑐𝑖𝑗𝑚 = 𝑐𝑗𝑖𝑚

Tensor anti-symmetric (or skew-symmetric)


𝑠𝑖𝑗 = −𝑠𝑗𝑖

𝑐𝑖𝑗𝑚 = −𝑐𝑗𝑖𝑚

08/07/2017 Dr. Luis Mosquera L. 43


Example:
Show that the inner product 𝑠𝑖𝑗 𝑎𝑖𝑗 of a symmetric tensor S and
an anti-symmetric tensor a is zero.

Solution:
𝑠𝑖𝑗 𝑎𝑖𝑗 = −𝑠𝑗𝑖 𝑎𝑗𝑖 = −𝑠𝑚𝑛 𝑎𝑚𝑛 = −𝑠𝑖𝑗 𝑎𝑖𝑗

Therefore,
2𝑠𝑖𝑗 𝑎𝑖𝑗 = 0

08/07/2017 Dr. Luis Mosquera L. 44


Example:

By direct expansion of the expression 𝑣𝑖 = 𝜀𝑖𝑗𝑘 𝑤𝑗𝑘 determine the components

of the vector Vi in terms of the components of the tensor Wjk

Solution:

By summing first on j and then on k and then omitting the zero terms, we find that

𝑣𝑖 = 𝜀𝑖1𝑘 𝑤1𝑘 + 𝜀𝑖2𝑘 𝑤2𝑘 + 𝜀𝑖3𝑘 𝑤3𝑘


𝑣𝑖 = 𝜀𝑖12 𝑤12 + 𝜀𝑖13 𝑤13 + 𝜀𝑖21 𝑤21 + 𝜀𝑖23 𝑤23 + 𝜀𝑖31 𝑤31 + 𝜀𝑖32 𝑤32

Therefore,

08/07/2017 Dr. Luis Mosquera L. 45


𝑣1 = 𝜀123 𝑤23 + 𝜀132 𝑤32 = 𝑤23 − 𝑤32

𝑣2 = 𝜀213 𝑤13 + 𝜀231 𝑤31 = 𝑤31 − 𝑤13

𝑣3 = 𝜀312 𝑤12 + 𝜀321 𝑤21 = 𝑤12 − 𝑤21

A skew-symmetric second order tensor 𝑊 = 𝑤 𝑒 𝑒 can be represented in terms of


𝑖𝑗 𝑖 𝑗

an axial vector by using the permutation symbol. Let the axial vector for Wij be wi defined by:
1
𝑤𝑖 = − 𝜀𝑖𝑗𝑘 𝑤𝑗𝑘
2
and wjk in terms of wi:
𝜀𝑖𝑚𝑛 𝑤𝑖 = −𝑤𝑚𝑛

08/07/2017 Dr. Luis Mosquera L. 46


Matrices and determinants
A matrix having elements Aij, which may be numbers, variables, functions, or any of
several mathematical entities, is designated by 𝐴𝑖𝑗

𝐴11 𝐴12 … 𝐴1𝑁


𝐴21 𝐴22 … 𝐴2𝑁
.
𝐴 = 𝐴𝑖𝑗 = ..

𝐴𝑀1 𝐴𝑀2 … 𝐴𝑀𝑁

Row and column matrices represent vectors, whereas a 3x3 square matrix
represents a second-order tensor.
For a diagonal matrix, Aij = 0 for i ≠ j
The unit or identity matrix I, is a diagonal matrix whose diagonal elements all have
the value one.

08/07/2017 Dr. Luis Mosquera L. 47


The NxM matrix formed by interchanging the rows and columns of the MxN matrix
A is called the transpose of A, and is written as 𝐴𝑇

Where, 𝐴𝑇𝑖𝑗 = 𝐴𝑗𝑖

A square matrix for which 𝐴 = 𝐴𝑇 is called a symmetric matrix

A square matrix for which 𝐴 = −𝐴𝑇 is called an anti-symmetric matrix

The elements of the principal diagonal of a skew-symmetric matrix are all zeros.

Two matrices are equal if they are identical element by element.

08/07/2017 Dr. Luis Mosquera L. 48


A+B = B+A
A+(B+C) = (A+B)+C
C = A B, CIJ = AIK BKJ
AB ≠ BA

𝐴𝑚 𝐴𝑛 = 𝐴𝑚 +𝑛
𝐴𝑛 𝑇
= 𝐴𝑇 𝑛

1
If BB = A, then: 𝐵 = 𝐴 = 𝐴2

Example: Use indicial notation to show that for arbitrary matrices A y B:


𝑇
𝐴𝐵 = 𝐵𝑇 𝐴𝑇

08/07/2017 Dr. Luis Mosquera L. 49


The determinant:

𝐴11 𝐴12 𝐴13


𝑑𝑒𝑡𝐴 = 𝐴𝑖𝑗 = 𝐴21 𝐴22 𝐴23
𝐴31 𝐴32 𝐴33

𝐴22 𝐴23 𝐴 𝐴23 𝐴 𝐴22


𝑑𝑒𝑡𝐴 = 𝐴11 − 𝐴12 21 + 𝐴13 21
𝐴32 𝐴33 𝐴31 𝐴33 𝐴31 𝐴32

𝑑𝑒𝑡𝐴 = 𝜀𝑖𝑗𝑘 𝐴𝑖1 𝐴𝑗2 𝐴𝑘3 = 𝜀𝑖𝑗𝑘 𝐴1𝑖 𝐴2𝑗 𝐴3𝑘

det 𝐴𝐵 = det 𝐵𝐴 = 𝑑𝑒𝑡𝐴 𝑑𝑒𝑡𝐵

08/07/2017 Dr. Luis Mosquera L. 50


Example: show that det A = det 𝐴𝑇

Solution:

Since
𝐴11 𝐴21 𝐴31
𝐴𝑇 = 𝐴12 𝐴22 𝐴32
𝐴13 𝐴23 𝐴33

Cofactor expansion by the first column here yields

𝐴22 𝐴32 𝐴 𝐴31 𝐴 𝐴31


𝑑𝑒𝑡𝐴𝑇 = 𝐴11 − 𝐴12 21 + 𝐴13 21 = det 𝐴
𝐴23 𝐴33 𝐴23 𝐴33 𝐴22 𝐴32

08/07/2017 Dr. Luis Mosquera L. 51


The inverse of the matrix A is written A-1

𝐴𝐴−1 = 𝐴−1 𝐴 = 𝐼

𝐴𝐵 = 𝐼 → 𝐵 = 𝐴−1 𝑎𝑛𝑑 𝐴 = 𝐵 −1

The adjoint matrix A* is defined as the transpose of the cofactor matrix

𝐴∗ = 𝐴𝑐 𝑇

The inverse matrix is expressed by

−1
𝐴∗
𝐴 =
det 𝐴

08/07/2017 Dr. Luis Mosquera L. 52


An orthogonal matrix, call it Q, is a square matrix for which:

𝑄 −1 = 𝑄 𝑇

Note that a symmetric orthogonal matrix is its own inverse

Also, if A and B are orthogonal matrices,


−1
𝐴𝐵 = 𝐵 −1 𝐴−1 = 𝐵 𝑇 𝐴𝑇 = 𝐴𝐵 𝑇

Problem 6:
If A is any orthogonal matrix, show that det A = ± 1

08/07/2017 Dr. Luis Mosquera L. 53


The vector-tensor product, u = T v or 𝑢𝑖 = 𝑇𝑖𝑗 𝑣𝑗 appears in matrix form as:

𝑢1 𝑇11 𝑇12 𝑇13 𝑣1


𝑢𝑖1 = 𝑇𝑖𝑗 𝑣𝑗 1 𝑜𝑟 𝑢2 = 𝑇21 𝑇22 𝑇23 𝑣2
𝑢3 𝑇31 𝑇32 𝑇33 𝑣3

In much the same way the product w = v. T or 𝑤𝑖 = 𝑣𝑗 𝑇𝑗𝑖


appears as:

𝑇11 𝑇21 𝑇31


𝑤1𝑖 = 𝑣1𝑗 𝑇𝑗 𝑖 𝑜𝑟 𝑤1 𝑤2 𝑤3 = 𝑣1 𝑣2 𝑣3 𝑇12 𝑇22 𝑇32
𝑇13 𝑇23 𝑇33

𝑤 = 𝑣. 𝑇 = (𝑇. 𝑣)𝑇 = 𝑣 𝑇 . 𝑇 𝑇 = 𝑢𝑇 = 𝑇. 𝑣

08/07/2017 Dr. Luis Mosquera L. 54


Problem 7. Show that the square matrices

1 0 0
5 2
𝐵𝑖𝑗 = 0 −1 0 and 𝐶𝑖𝑗 =
−12 −5
0 0 1
are both square roots of the identity matrix

08/07/2017 Dr. Luis Mosquera L. 55


Transformations of Cartesian Tensors

Transformation table between 𝑂𝑥1 𝑥2 𝑥3 𝑎𝑛𝑑 𝑂𝑥1′ 𝑥2′ 𝑥3′

𝑒1 , 𝑥1 𝑒2 , 𝑥2 𝑒3 , 𝑥3
𝑒1′ 𝑥1′ 𝑎11 𝑎12 𝑎13
𝑒2′ 𝑥2′ 𝑎21 𝑎22 𝑎23
𝑒3′ 𝑥3′ 𝑎31 𝑎32 𝑎33

𝑎𝑖𝑗 = 𝐶𝑜𝑠(𝑥𝑖′ , 𝑥𝑗 )

𝑒𝑖′ = 𝑎𝑖𝑗 𝑒𝑗
𝑒𝑖 = 𝑎𝑗 𝑖 𝑒𝑗′

08/07/2017 Dr. Luis Mosquera L. 56


𝑒1′ = 𝑎11 𝑒1 + 𝑎12 𝑒2 + 𝑎13 𝑒3 = 𝑎1𝑗 𝑒𝑗

𝑒2′ = 𝑎21 𝑒1 + 𝑎22 𝑒2 + 𝑎23 𝑒3 = 𝑎2𝑗 𝑒𝑗

𝑒3′ = 𝑎31 𝑒1 + 𝑎32 𝑒2 + 𝑎33 𝑒3 = 𝑎3𝑗 𝑒𝑗

In matrix form:

𝑒1′ 𝑎11 𝑎12 𝑎13 𝑒1



𝑒𝑖1 = 𝑎𝑖𝑗 𝑒𝑗 1 𝑜𝑟 𝑒2′ = 𝑎21 𝑎22 𝑎23 𝑒2
𝑒3′ 𝑎31 𝑎32 𝑎33 𝑒3

Transformation matrix

08/07/2017 Dr. Luis Mosquera L. 57


Because of the perpendicularity of the primed axes,

𝑒𝑖′ . 𝑒𝑗′ = 𝑎𝑖𝑞 𝑒𝑞 . 𝑎𝑗𝑚 𝑒𝑚 = 𝑎𝑖𝑞 𝑎𝑗𝑚 𝛿𝑞𝑚 = 𝑎𝑖𝑞 𝑎𝑗𝑞 = 𝛿𝑖𝑗

𝑎𝑖𝑞 𝑎𝑗𝑞 = 𝛿𝑖𝑗 𝑜𝑟 𝐴𝐴𝑇 = 𝐼

Consider next an arbitrary vector v having components vi in the unprimed system, and
vi´ in the primed system, then:
𝑣 ′ = 𝐴𝑣 = 𝑣𝐴𝑇
𝑣𝑗′ = 𝑎𝑗𝑖 𝑣𝑖

They are not the components of a second order cartesian tensor

08/07/2017 Dr. Luis Mosquera L. 58


The unprimed components in terms of the primed:

𝑣 = 𝑣 ′ 𝐴 = 𝐴𝑇 𝑣 ′

𝑣𝑘 = 𝑎𝑗𝑘 𝑣𝑗′

To the dyad uv we have:

𝑢𝑖 𝑣𝑗 = 𝑎𝑞𝑖 𝑢𝑞′ 𝑎𝑚𝑗 𝑣𝑚′ = 𝑎𝑞𝑖 𝑎𝑚𝑗 𝑢𝑞′ 𝑣𝑚′

A dyad is, after all, one form of a second-order tensor, then the transformation
law for a second-order tensor, T is:

𝑡𝑖𝑗 = 𝑎𝑞𝑖 𝑎𝑚𝑗 𝑡𝑞𝑚 𝑜𝑟 𝑇 = 𝐴𝑇 𝑇 ′ 𝐴

𝑡𝑖𝑗′ = 𝑎𝑖𝑞 𝑎𝑗 𝑚 𝑡𝑞𝑚 𝑜𝑟 𝑇 ′ = 𝐴𝑇𝐴𝑇

08/07/2017 Dr. Luis Mosquera L. 59


Rotation and reflection of reference axes

The primed axes may be related to the unprimed axes through either a rotation
about an axis through the origin, or by a reflection of the axes in one of the
coordinate planes, (or by a combination of such changes).

Example: Consider a 90° (counterclockwise) rotation about the x2 axis in figure:

0 0 −1
𝑎𝑖𝑗 = 0 1 0
1 0 0

det A = -1

proper orthogonal transformation

08/07/2017 Dr. Luis Mosquera L. 60


Example: For a reflection of axes in the x2 x3 plane shown in figure:

−1 0 0
𝑎𝑖𝑗 = 0 1 0
0 0 1

det A = -1

Improper orthogonal transformation

08/07/2017 Dr. Luis Mosquera L. 61


Principal values and principal directions of symmetric second-
order tensors

The inner product of a second order tensor T with a arbitrary vector u:

𝑡𝑖𝑗 𝑢𝑗 = 𝑣𝑖 𝑜𝑟 𝑇. 𝑢 = 𝑣

Any second-order tensor may be thought of as a linear transformation which


transforms the antecedent vector u into the image vector v in a Euclidean three-
space.

If: 𝑡𝑖𝑗 𝑢𝑗 = λ𝑢𝑖 𝑜𝑟 𝑇. 𝑢 = λ𝑢

Then, the direction defined by ui is called a principal direction, or eigenvector of


T, and the scalar λ is called a principal value, or eigenvalue of T.

08/07/2017 Dr. Luis Mosquera L. 62


Using the kronecker delta:

𝑡𝑖𝑗 − λ𝛿𝑖𝑗 𝑢𝑗 = 0 𝑜𝑟 𝑇 − λ𝐼 . 𝑢 = 0

This equations have non-trivial solutions only if the determinant of coeficients


vanishes. Thus: 𝑡 − λ𝛿 = 0
𝑖𝑗 𝑖𝑗

Which upon expansion leads to the cubic in λ (called the characteristic equation)

𝜆3 − 𝐼𝑇 𝜆2 + 𝐼𝐼𝑇 λ − 𝐼𝐼𝐼𝑇 = 0
Where,
𝐼𝑇 = 𝑡𝑖𝑖 = 𝑡𝑟 𝑇
1 1 2
𝐼𝐼𝑇 = (𝑡𝑖𝑖 𝑡𝑗𝑗 − 𝑡𝑖𝑗 𝑡𝑗𝑖 ) = 𝑡𝑟 𝑇 − 𝑡𝑟(𝑇 2 )
2 2

𝐼𝐼𝐼𝑇 = 𝜀𝑖𝑗𝑘 𝑡1𝑖 𝑡2𝑗 𝑡3𝑘 = det 𝑇

08/07/2017 Dr. Luis Mosquera L. 63


I, II, III, are known as the first, second and third invariants of the tensor T.

With each of the roots λq we can determine a principal direction 𝑛𝑖(𝑞)

𝑡𝑖𝑗 − λ𝑞 𝛿𝑖𝑗 𝑛𝑖𝑞 = 0

with 𝑛𝑖𝑞 . 𝑛𝑖𝑞 = 1 (𝑞 = 1,2,3)

If the λq are distint, the principal directions are unique and mutually
perpendicular. If, there is a pair of equal roots, say λ1 = λ2 , then only the direction
associated with λ3 will be unique. In this case any other two directions which are
orthogonal to 𝑛𝑖(3) and to one another so as to form a right handed system, may be
taken as principal directions.
If λ1 = λ2 = λ3 , every set of right-handed orthogonal axes qualifies as principal axes,
and every direction is said to be a principal direction.

08/07/2017 Dr. Luis Mosquera L. 64


Let the components of the tensor T be given initially with repect to arbitrary
cartesian axes , 𝑂𝑥1 𝑥2 𝑥3 and let the principal axes of T be designated by 𝑂𝑥1∗ 𝑥2∗ 𝑥3∗ ,
as shown in the figure:

08/07/2017 Dr. Luis Mosquera L. 65


Principal axes 𝑂𝑥1∗ 𝑥2∗ 𝑥3∗ relative to axes 𝑂𝑥1 𝑥2 𝑥3

𝑥1 𝑜𝑟 𝑒1 𝑥2 𝑜𝑟 𝑒2 𝑥3 𝑜𝑟 𝑒3
𝑥1∗ 𝑜𝑟 𝑒1∗ 𝑎11 = 𝑛1
(1)
𝑎12 = 𝑛2
(1)
𝑎13 = 𝑛3
(1)

𝑥2∗ 𝑜𝑟 𝑒2∗ (2)


𝑎21 𝑛1 𝑎22 = 𝑛2
(2)
𝑎23 = 𝑛3
(2)

𝑥3∗ 𝑜𝑟 𝑒3∗ (3)


𝑎31 𝑛1 𝑎32 = 𝑛2
(3)
𝑎33 = 𝑛3
(3)

The transformation matrix here is orthogonal and in accordance with the


transformation law for second-order tensors:

𝑡𝑖𝑗∗ = 𝑎𝑖𝑞 𝑎𝑗𝑚 𝑡𝑞𝑚 𝑜𝑟 𝑇 ∗ = 𝐴𝑇𝐴𝑇

Where T* is a diagonal matrix whose elements are the principal values λ(q)

08/07/2017 Dr. Luis Mosquera L. 66


Example:

Determine the principal values and principal directions of the second-order


tensor T whose matrix representation is:

5 2 0
𝑡𝑖𝑗 = 2 2 0
0 0 3

Solution:

5−𝜆 2 0
2 2−𝜆 0 =0
0 0 3−𝜆

3 − λ 10 − 7λ + λ2 − 4 = 0

For λ1=3: n1 = n2 = 0 and so from nini =1 we have n3 = ± 1

08/07/2017 Dr. Luis Mosquera L. 67


For λ2 = 6: n3 = 0; n1= ± 2/ 5; n2 = ±1/ 5

For λ3 =1: n3 = 0; n1= ± 1/ 5; n2 = ±2/ 5

From these results the transformation matrix A is given by:

0 0 ±1
2 1
± ± 0
𝑎𝑖𝑗 = 5 5
1 2
± ± 0
5 5

Finally, 0 0 1 2 1
2 1 0
0 5 2 0 5 5 3 0 0
5 5 2 2 0 1 2 = 0 6 0
1 2 0 0 3 0 5

5 0 0 1
− 0
5 5 1 0 0

08/07/2017 Dr. Luis Mosquera L. 68


Example: Determine the principal values and principal directions of the second-
order tensor T whose matrix representation is:

5 1 2
1 5 2
2 2 6

08/07/2017 Dr. Luis Mosquera L. 69


Tensor Fields, Tensor Calculus

A tensor field assigns to every location x, at every instant of time t, a tensor tij…k(x,t),for
which x ranges over a finite region of space, and t varies over some interval of time.

∅ 𝑥, 𝑡 Scalar field
𝑣𝑖 (𝑥, 𝑡) Vector field
𝑡𝑖𝑗 (𝑥, 𝑡) Tensor field

Partial differentiation of a tensor field with respect to the variable t is symbolized by


𝜕
the operator 𝜕𝑡 and follows the usual rules of calculus. Partial differentiation with
𝜕
respect to the coordinate 𝑥𝑞 will be indicated by the operator 𝜕𝑥 , which may be
𝑞
abbreviated as simply 𝜕𝑞 .

08/07/2017 Dr. Luis Mosquera L. 70


𝜕2
Likewise, the second partial may be written 𝜕𝑞𝑚 , and so on.
𝜕𝑥𝑞 𝜕𝑥𝑚

𝜕∅
∅,𝑖 =
𝜕𝑥𝑖

𝜕𝑣𝑖
𝑣𝑖,𝑗 =
𝜕𝑥𝑗

𝜕𝑡𝑖𝑗
𝑡𝑖𝑗,𝑘 =
𝜕𝑥𝑘

Note that differentiation with respect to a coordinate produces a tensor of one


order higher.
𝜕 2 𝑢𝑖
𝜕𝑥𝑗 𝜕𝑥𝑘

08/07/2017 Dr. Luis Mosquera L. 71


Also,

𝜕𝑥𝑖
= 𝛿𝑖𝑗
𝜕𝑥𝑗

𝜕 𝜕 𝜕 𝜕
𝛻= 𝑒 + 𝑒2 + 𝑒 = 𝑒 = 𝜕𝑖
𝜕𝑥1 1 𝜕𝑥2 𝜕𝑥3 3 𝜕𝑥𝑖 𝑖

𝛻∅ = 𝑔𝑟𝑎𝑑 ∅ = 𝜕𝑖 ∅ = ∅,𝑖

𝛻𝑣 = 𝜕𝑖 𝑣𝑗 = 𝑣𝑗,𝑖

𝛻. 𝑣 = 𝜕𝑖 𝑣𝑖 = 𝑣𝑖,𝑖

𝛻 × 𝑣 = 𝜀𝑖𝑗𝑘 𝜕𝑗 𝑣𝑘 𝜀𝑖𝑗𝑘 𝑣𝑘,𝑗

08/07/2017 Dr. Luis Mosquera L. 72


Example:
Show that div(curl v) = 0

Solution

For any vector v we write 𝜕𝑖 𝜀𝑖𝑗𝑘 𝑣𝑘,𝑗 = 𝜀𝑖𝑗𝑘 𝑣𝑘,𝑗𝑖 = 0

And because the first term of this inner product is skew-symmetric in i and j, whereas
the second term is symmetric in the same indices, (since vk is assumed to have
continuos spatial gradients), their product is zero.

08/07/2017 Dr. Luis Mosquera L. 73


Problem 8. Show that the square matrices are both square roots of the identity matrix.

1 0 0
𝐵𝑖𝑗 = 0 −1 0
0 0 1
5 2
𝐶𝑖𝑗 =
−12 −5

Problem 9. The angles between the respective axes of the 𝑂𝑥1∗ 𝑥2∗ 𝑥3∗ and the 𝑂𝑥1 𝑥2 𝑥3

cartesian systems are given by the table below

𝑥1 𝑥2 𝑥3
𝑥1′ 450 900 450
𝑥2′ 600 450 1200
𝑥3′ 1200 450 600

Determine the transformation matrix between the two sets of axes, and show that it is
a proper orthogonal transformation.

08/07/2017 Dr. Luis Mosquera L. 74


Stress Principles

In continuum mechanics a body is considered stress


free if the only forces present are those inter-atomic
forces required to hold the body together.

08/07/2017 Dr. Luis Mosquera L. 75


Body and surface forces,mass density

Those forces acting on all volume elements, and are distributed


throughout the body, are known as body forces:
bi (force per unit mass)
pi (force per unit volume)

Those forces which act upon, and are distributed in some fashion
over a surface element of the body, regardless of whether that
element is part of the bounding surface, or an arbitrary element of
surface within the body, are called surface forces fi.

08/07/2017 Dr. Luis Mosquera L. 76


Density

∆𝑚
𝜌𝑎𝑣𝑒 =
∆𝑣
𝑑𝑚
𝜌=
𝑑𝑣

𝜌b = p

Of course, the density is, in general, a scalar function of position and time as
indicated by: 𝜌 = 𝜌 𝑥𝑖 , 𝑡 , and thus may vary from point to point within a
given body.

08/07/2017 Dr. Luis Mosquera L. 77


Cauchy Stress Principle
Consider a homogeneous, isotropic material body having a bounding surface
S, and a volume V, which is subjected to arbitrary surface forces fi and body
forces bi. Let P be an interior point and a plane surface 𝑆 ∗ passing through
point P (cutting plane).

08/07/2017 Dr. Luis Mosquera L. 78


The Cauchy stress principle asserts that in the limit as the area 𝑆 ∗ shrinks
to zero with P remaining an interior point, we obtain:

∆𝑓𝑖 𝑑𝑓𝑖 (𝑛 )
lim = = 𝑡𝑖
∆𝑆 ∗ →0 ∆𝑆 ∗ 𝑑𝑆 ∗

and
∆𝑀𝑖
lim =0
∆𝑆 ∗ →0 ∆𝑆 ∗

The vector 𝑡𝑖 𝑛 is called the stress vector, or sometimes the traction vector

For the infinity of cutting planes imaginable through point P, each identified
by a specific 𝑛, there is also an infinity of associated stress vector 𝑡𝑖 𝑛 for a
given loading of the body. The totality of pairs 𝑡𝑖 𝑛 defines the state of stress
at that point.
08/07/2017 Dr. Luis Mosquera L. 79
We have made the assumption that in the limit at P, the moment vector
vanishes, and there is no remaining concentrated moment, or couple stress
as it is called.

By applying Newton’s law:

(𝑛 ) 𝑑
𝑡𝑖 𝑑𝑆 + 𝜌𝑏𝑖 𝑑𝑣 = 𝜌𝑣𝑖 𝑑𝑣
𝑆𝐼 𝑣𝐼 𝑑𝑡 𝑣𝐼

(𝑛 ) 𝑑
𝑡𝑖 𝑑𝑆 + 𝜌𝑏𝑖 𝑑𝑣 = 𝜌𝑣𝑖 𝑑𝑣
𝑆𝐼𝐼 𝑣𝐼𝐼 𝑑𝑡 𝑣𝐼𝐼

Where SI y SII are the bounding surfaces and VI and VII are the volumes of
portions I and II, respectively. Also, vi is the velocity field for the two portions.

08/07/2017 Dr. Luis Mosquera L. 80


The linear momentum principle may also be applied to the body as a whole,
so that:

(𝑛 ) 𝑑
𝑡𝑖 𝑑𝑆 + 𝜌𝑏𝑖 𝑑𝑣 = 𝜌𝑣𝑖 𝑑𝑣
𝑆 𝑣 𝑑𝑡 𝑣

Where, (𝑛 ) (−𝑛 )
𝑡𝑖 + 𝑡𝑖 𝑑𝑆 = 0
𝑆∗

(𝑛 ) (−𝑛 )
Which means that: 𝑡𝑖 = −𝑡𝑖

08/07/2017 Dr. Luis Mosquera L. 81


The Stress Tensor

The Cauchy stress principle associates with each direction 𝑛 at point P a stress
vector 𝑡𝑖 𝑛 :

These three stress vectors associated with the coordinate planes are expressed by:

(𝑒 𝑖 ) (𝑒 𝑖 )
𝑡 = 𝑡𝑗 𝑒𝑗

08/07/2017 Dr. Luis Mosquera L. 82


For deducir an expression for the coordinate components of the stress
vector at P associated with an arbitrarily oriented plane, we consider the
equilibrium of a small portion of the body:

𝑛 = 𝑛𝑖 𝑒𝑖
𝑛 is perpendicular to the base ABC

08/07/2017 Dr. Luis Mosquera L. 83


If the area of the base is assigned the value dS, the areas of the respective
faces will be the projected areas:
𝑒𝑖
𝑑𝑆𝑖 = 𝑑𝑆 cos⁡
(𝑛, 𝑒𝑖 )
𝑛
for BPC 𝑑𝑆1 = 𝑛1 𝑑𝑆

for CPA 𝑑𝑆2 = 𝑛2 𝑑𝑆

for APB 𝑑𝑆3 = 𝑛3 𝑑𝑆

Equilibrium requires the vector sum of all forces acting on the tetrahedron to
be zero, that is, for,

∗ (𝑛 ) 𝑒1 𝑒2 𝑒3
𝑡𝑖 𝑑𝑆 − ∗𝑡𝑖 𝑑𝑆1 − ∗𝑡𝑖 𝑑𝑆2 − ∗𝑡𝑖 𝑑𝑆3 + 𝜌∗ 𝑏𝑖 𝑑𝑣 = 0

∗ (𝑛 ) ∗ 𝑒𝑗
𝑡𝑖 𝑑𝑆 − 𝑡𝑖 𝑛𝑗 𝑑𝑆 + 𝜌∗ 𝑏𝑖 𝑑𝑣 = 0
08/07/2017 Dr. Luis Mosquera L. 84
∗ (𝑛 ) ∗ 𝑒𝑗
𝑡𝑖 𝑑𝑆 − 𝑡𝑖 𝑛𝑗 𝑑𝑆 + 𝜌∗ 𝑏𝑖 𝑑𝑣 = 0

The volume of the tetrahedron is given by


1
𝑑𝑣 = 3 h dS

∗ (𝑛 ) ∗ 𝑒𝑗 1 ∗
Where, 𝑡𝑖 = 𝑡𝑖 𝑛𝑗 − 𝜌 𝑏𝑖 ℎ
3

Taking the limit as h → 0


(𝑛 ) 𝑒𝑗
𝑡𝑖 = 𝑡𝑖 𝑛𝑗

08/07/2017 Dr. Luis Mosquera L. 85


𝑒𝑗
defining 𝑡𝑗𝑖 ≡ 𝑡𝑖 , then,

(𝑛 )
𝑡𝑖 = 𝑡𝑗𝑖 𝑛𝑗 𝑜𝑟 𝑡 (𝑛 ) = 𝑛. 𝑇

Stress as a Tensor
𝑒𝑗
The quantities 𝑡𝑗𝑖 ≡ 𝑡𝑖
are the components of a second order tensor T know as the stress
tensor.

Since 𝑡 𝑛 can be expressed in terms of its components in either coordinate system,

𝑡 (𝑛 ) = 𝑡𝑗𝑖 𝑛𝑗 𝑒𝑖 = 𝑡𝑗𝑖′ 𝑛𝑗′ 𝑒𝑖′

08/07/2017 Dr. Luis Mosquera L. 86


as, 𝑒𝑖′ = 𝑎𝑖𝑗 𝑒𝑗 , and, 𝑛𝑗′ = 𝑎𝑗 𝑠 𝑛𝑠

Then, 𝑡𝑠𝑟 − 𝑎𝑗𝑠 𝑎𝑖𝑟 𝑡𝑗𝑖′ 𝑛𝑠 𝑒𝑟 = 0

Because the vectors 𝑒𝑟 are linearly independent, we see that:

𝑡𝑠𝑟 = 𝑎𝑗𝑠 𝑎𝑖𝑟 𝑡𝑗𝑖′

That is the transformation equation for a second-order tensor

Also, we see that if know the stress vectors on the three coordinate planes
of any cartesian system at P, or equivalently, the nine stress tensor
components tji at that point, we can determine the stress vector for any
plane at that point.

08/07/2017 Dr. Luis Mosquera L. 87


𝑡 (𝑛 ) = 𝑛. 𝑇
In the matrix form:
𝑡11 𝑡12 𝑡13
(𝑛 ) (𝑛 ) (𝑛 )
𝑡1 𝑡2 𝑡3 = 𝑛1 𝑛2 𝑛3 𝑡21 𝑡22 𝑡23
𝑡31 𝑡32 𝑡33

The nine components of tij are often displayed by arrows on the coordinate
faces of a rectangular parallelpiped, as shown in the figure:

08/07/2017 Dr. Luis Mosquera L. 88


Example: Let the components of the stress tensor at P be given in matrix form by

21 −63 42
𝑡𝑖𝑗 = −63 0 84
42 84 −21

In units of mega-Pascals. Determine:

(a) The stress vector on the plane at P having the unit normal
1
𝑛= (2𝑒1 − 3𝑒2 + 6𝑒3 )
7

(b) The stress vector on a plane at P parallel to the plane ABC


shown in the sketch

08/07/2017 Dr. Luis Mosquera L. 89


Solution

(a)

(𝑛 ) (𝑛 ) (𝑛 ) 2 3 6 21 −63 42
𝑡1 𝑡2 𝑡3 = − −63 0 84 = 69 54 −42
7 7 7 42 84 −21

(b) The equation of the plane ABC is 2𝑥1 + 2𝑥2 + 𝑥3 =2, and the unit outward
1
normal to this plane is 𝑛= 2𝑒1 + 2𝑒2 + 𝑒3 so that,
3

(𝑛 ) (𝑛 ) (𝑛 ) 2 2 1 21 −63 42
𝑡1 𝑡2 𝑡3 = −63 0 84 = −14 −14 77
3 3 3 42 84 −21

08/07/2017 Dr. Luis Mosquera L. 90


Stress Tensor Symmetry

Consider a material body having a volume V and a bounding surface S. Let the
body be subjected to surface tractions 𝑡𝑖(𝑛 ) and body forces bi, as shown by
figure:

As before, we exclude concentrated body moments from consideration.

08/07/2017 Dr. Luis Mosquera L. 91


Equilibrium requires that

(𝑛 )
𝑡𝑖 𝑑𝑆 + 𝜌𝑏𝑖 𝑑𝑣 = 0
𝑆 𝑣

(𝑛 )
𝑡𝑖 𝑑𝑆 = 𝑡𝑗𝑖 𝑛𝑗 𝑑𝑆 = 𝑡𝑗𝑖 ,𝑗 𝑑𝑣
𝑆 𝑆 𝑣

So that

(𝑡𝑗𝑖 ,𝑗 + 𝜌𝑏𝑖 )𝑑𝑣 = 0


𝑣

This equation must be valid for an arbitrary volume V

→ 𝑡𝑗𝑖 ,𝑗 + 𝜌𝑏𝑖 = 0

08/07/2017 Dr. Luis Mosquera L. 92


In addition to the balance of forces, equilibrium requires that the summation
of moments with respect to an arbitrary point must also be zero. Therefore,
taking the origin of coordinates as the center for moments, express the
balance of moments for the body as a whole by

(𝑛 )
𝜀𝑖𝑗𝑘 𝑥𝑗 𝑡𝑖 𝑑𝑆 + 𝜀𝑖𝑗𝑘 𝑥𝑗 𝜌𝑏𝑖 𝑑𝑣 = 0
𝑆 𝑣

(𝑛 )
As before, using 𝑡𝑘 = 𝑡𝑞𝑘 𝑛𝑞 and Gauss’s divergence theorem, obtain

𝜀𝑖𝑗𝑘 𝑥𝑗 ,𝑞 𝑡𝑞𝑘 + 𝑥𝑗 (𝑡𝑞𝑘 ,𝑞 + 𝜌𝑏𝑘 ) 𝑑𝑣 = 0


𝑣

But,
𝑥𝑗 ,𝑞 = 𝛿𝑗𝑞 and 𝑡𝑘𝑞 ,𝑘 + 𝜌𝑏𝑘 = 0

08/07/2017 Dr. Luis Mosquera L. 93


Then,

𝜀𝑖𝑗𝑘 𝑡𝑗 𝑘 𝑑𝑣 = 0
𝑣

Again, since volume V is arbitrary,

𝜀𝑖𝑗𝑘 𝑡𝑗 𝑘 = 0

By a direct expansion of this equation, we obtain

𝑡𝑗 𝑘 = 𝑡𝑘𝑗

08/07/2017 Dr. Luis Mosquera L. 94


Stress transformation laws

𝑡𝑖𝑗′ = 𝑎𝑖𝑞 𝑡𝑞𝑚 𝑎𝑗𝑚 𝑜𝑟 𝑇 ′ = 𝐴𝑇𝐴𝑇

08/07/2017 Dr. Luis Mosquera L. 95


Example: Let the stress components (in Mpa) at point P with respect to axes P
X1X2X3 be expressed by the matrix

1 3 2
𝑡𝑖𝑗 = 3 1 0
2 0 −2

And let the primed axes 𝑃𝑥 ′ 𝑥 ′ 𝑥 ′ be obtained by a 45° counterclockwise


1 2 3

rotation about the x3 axis. Determine the stress components 𝑡𝑖𝑗′

08/07/2017 Dr. Luis Mosquera L. 96


solution

𝑐𝑜𝑠𝜃 𝑠𝑖𝑛𝜃 0
𝑎𝑖𝑗 = −𝑠𝑖𝑛𝜃 𝑐𝑜𝑠𝜃 0
0 0 1

𝑇 ′ = 𝐴𝑇𝐴𝑇

4 0 2
𝑡𝑖𝑗′ = 0 −2 − 2
2 − 2 −2

08/07/2017 Dr. Luis Mosquera L. 97


Principal Stresses; Principal Stress Directions

For certain special directions at P, the stress vector does indeed act in the
direction of ni , thus, for such directions
(𝑛 )
𝑡𝑖 = 𝜎 𝑛𝑖

Directions designated by ni are called principal stress directions, and the


scalar σ is called a principal stress value of tij. Formulating the eigenvalue
problem for the stress tensor

𝑡𝑖𝑗 𝑛𝑗 − 𝜎𝛿𝑖𝑗 𝑛𝑗 = 0

08/07/2017 Dr. Luis Mosquera L. 98


For non-trivial solutions

𝑡𝑖𝑗 − 𝛿𝑖𝑗 𝜎 = 0

𝑥1 𝑥2 𝑥3
𝑥1∗ 𝑎11 = 𝑛1
(1)
𝑎12 = 𝑛2
(1) (1)
𝑎13 = 𝑛3
𝑥2∗ (2)
𝑎21 = 𝑛1
(2)
𝑎22 = 𝑛2
(2)
𝑎23 = 𝑛3
𝑥3∗ (3)
𝑎31 = 𝑛1
(3)
𝑎32 = 𝑛2
(3)
𝑎33 = 𝑛3
08/07/2017 Dr. Luis Mosquera L. 99
Example: The components of the stress tensor at P are given in MPa
with respect to axes P X1 X2 X3 by the matrix

57 0 24
𝑡𝑖𝑗 = 0 50 0
24 0 43
determine the principal stresses and the principal stress directions at P

08/07/2017 Dr. Luis Mosquera L. 100


Solution:
4 5 0 3 5
A= 0 1 0
−3 5 0 4 5

4 5 0 3 5 57 0 24 4 5 0 −3 5
𝑇′ = 0 1 0 . 0 50 0 . 0 1 0
−3 5 0 4 5 24 0 43 3 5 0 4 5

75 0 0
𝑇′ = 0 50 0
0 0 25

08/07/2017 Dr. Luis Mosquera L. 101


Maximum and Minimum Stress Values
(𝑛 )
The stress vector 𝑡𝑖 on an arbitrary plane at P may be resolved into a component
normal to the plane having a magnitude 𝜎𝑁 , along with a shear component which acts
in the plane and has a magnitude 𝜎𝑆

𝜎𝑁 is given by the dot product, 𝜎𝑁 = 𝑡 (𝑛 ) 𝑛𝑖


𝑖

(𝑛 )
As, 𝑡𝑖 = 𝑡𝑖𝑗 𝑛𝑗

It follows that, 𝜎𝑁 = 𝑡𝑖𝑗 𝑛𝑖 𝑛𝑗 𝑜𝑟 𝜎𝑁 = 𝑡 (𝑛 ) . 𝑛

𝜎𝑆2 = 𝑡 (𝑛 ) . 𝑡 (𝑛 ) − 𝜎𝑁2

08/07/2017 Dr. Luis Mosquera L. 102


We may use the Lagrangian multiplier method to obtain extremal values of 𝜎𝑁

𝑓 𝑛𝑖 = 𝑡𝑖𝑗 𝑛𝑖 𝑛𝑗 − 𝜎(𝑛𝑖 𝑛𝑖 − 1)

The method requires that,

𝜕𝑓
= 𝑡𝑖𝑗 𝛿𝑖𝑘 𝑛𝑗 + 𝛿𝑗 𝑘 𝑛𝑖 − 𝜎 2𝑛𝑖 𝛿𝑖𝑘 = 0
𝜕𝑛𝑘

This equation reduces to


𝑡𝑖𝑗 − 𝜎𝛿𝑖𝑗 𝑛𝑗 = 0

Therefore , we conclude that the Lagrangian multiplier σ assumes the role of a


principal stress and, that the principal stresses include both the maximum and
minimum normal stress values.

08/07/2017 Dr. Luis Mosquera L. 103


Let the principal stresses be ordered in the sequence σI > σII > σIII so that

𝑡 (𝑛 ) = 𝑇. 𝑛 = 𝜎𝐼 𝑛1 𝑒1∗ + 𝜎𝐼𝐼 𝑛2 𝑒2∗ + 𝜎𝐼𝐼𝐼 𝑛3 𝑒3∗

𝜎𝑁 = 𝑡 (𝑛 ) . 𝑛 = 𝜎𝐼 𝑛12 + 𝜎𝐼𝐼 𝑛22 + 𝜎𝐼𝐼𝐼 𝑛32

𝜎𝑆2 = 𝑡 (𝑛 ) . 𝑡 (𝑛 ) − 𝜎𝑁2
𝜎𝑆2 = 𝜎𝐼2 𝑛12 + 𝜎𝐼𝐼2 𝑛22 +𝜎𝐼𝐼𝐼
2 2
𝑛3 − 𝜎𝐼 𝑛12 + 𝜎𝐼𝐼 𝑛22 + 𝜎𝐼𝐼𝐼 𝑛32 2

𝑛32 = 1 − 𝑛12 − 𝑛22



2
𝜎𝑆2 = 𝜎𝐼 − 𝜎𝐼𝐼𝐼 𝑛12 + 𝜎𝐼𝐼2 − 𝜎𝐼𝐼𝐼
2
𝑛22 + 𝜎𝐼𝐼𝐼
2
− 𝜎𝐼 − 𝜎𝐼𝐼𝐼 𝑛12 + 𝜎𝐼𝐼 − 𝜎𝐼𝐼𝐼 𝑛22 + 𝜎𝐼𝐼𝐼

To obtain the extremal values of 𝜎𝑆2 , we must equate the derivatives of the right-hand
side of this equation with respect to both n1 and n2 to zero, and solve simultaneously.

08/07/2017 Dr. Luis Mosquera L. 104


Obtaining the extremal values:

1 1 1
𝑛1 = 0, 𝑛2 = ± , 𝑛3 = ± ; 𝜎𝑆 = 𝜎𝐼𝐼 − 𝜎𝐼𝐼𝐼
2 2 2

1 1 1
𝑛1 = ± , 𝑛2 = 0, 𝑛3 = ± ; 𝜎𝑆 = 𝜎𝐼𝐼𝐼 − 𝜎𝐼
2 2 2

1 1 1
𝑛1 = ± , 𝑛2 = ± , 𝑛3 = 0; 𝜎𝑆 = 𝜎𝐼 − 𝜎𝐼𝐼
2 2 2

It is clear that because 𝜎𝐼 > 𝜎𝐼𝐼 > 𝜎𝐼𝐼𝐼 , the largest shear stress value is

1
𝜎𝑆𝑚𝑎𝑥 = 𝜎𝐼𝐼𝐼 − 𝜎𝐼
2

08/07/2017 Dr. Luis Mosquera L. 105


Example: The state of stress at point P is given in MPa with respect to axes X1X2X3
by the matrix

25 0 0
𝑡𝑖𝑗 = 0 −30 −60
0 −60 5

(a) Determine the stress vector on the plane whose unit normal is
1
𝑛= 2𝑒1 + 𝑒2 + 2𝑒3
3

(b) Determine the normal stress component 𝜎𝑁 and shear component 𝜎𝑆


on the same plane.

08/07/2017 Dr. Luis Mosquera L. 106


Solution:

(a) 2
(𝑛)
3
𝑡1 25 0 0
(𝑛) 1 1 50
𝑡2 = 𝑡𝑖𝑗 𝑛𝑗 = 0 −30 −60 3 = −150
3
(𝑛) 0 −60 5 −50
𝑡3 2
3

(b) 2
3
2 1 2 25 0 0 1 150
𝜎𝑁 = 0 −30 −60 3 = − 𝑀𝑃𝑎
3 3 3 0 −60 9
5
2
3

𝜎𝑆2 = 𝑡 (𝑛 ) . 𝑡 (𝑛 ) − 𝜎𝑁2 = 2777 𝑀𝑃𝑎2

08/07/2017 Dr. Luis Mosquera L. 107


Mohr’s Circles for Stress

It provides a graphic means for the transformation of a second order tensor. In


three dimensions, the convenience and simplicity of Mohr’s circle to transform
stress components to an arbitrary reference frame vanishes.

Consider the state of stress at P


referenced to principal axes:

𝑡 (𝑛 ) = 𝑇. 𝑛 = 𝜎𝐼 𝑛1 𝑒1∗ + 𝜎𝐼𝐼 𝑛2 𝑒2∗ + 𝜎𝐼𝐼𝐼 𝑛3 𝑒3∗

𝜎𝑁 = 𝑡 (𝑛 ) . 𝑛 = 𝜎𝐼 𝑛12 + 𝜎𝐼𝐼 𝑛22 + 𝜎𝐼𝐼𝐼 𝑛32

𝜎𝑁2 + 𝜎𝑆2 = 𝜎𝐼2 𝑛12 + 𝜎𝐼𝐼2 𝑛22 + 𝜎𝐼𝐼𝐼


2 2
𝑛3

𝑛12 + 𝑛22 + 𝑛32 = 1

08/07/2017 Dr. Luis Mosquera L. 108


Solving these equations, we obtain

𝜎𝑁 − 𝜎𝐼𝐼 𝜎𝑁 − 𝜎𝐼𝐼𝐼 + 𝜎𝑆2


𝑛12 =
𝜎𝐼 − 𝜎𝐼𝐼 𝜎𝐼 − 𝜎𝐼𝐼𝐼
2
𝜎𝑁 − 𝜎𝐼𝐼𝐼 𝜎𝑁 − 𝜎𝐼 + 𝜎𝑆
𝑛22 =
𝜎𝐼𝐼 − 𝜎𝐼𝐼𝐼 𝜎𝐼𝐼 − 𝜎𝐼

𝜎𝑁 − 𝜎𝐼 𝜎𝑁 − 𝜎𝐼𝐼 + 𝜎𝑆2
𝑛32 =
𝜎𝐼𝐼𝐼 − 𝜎𝐼 𝜎𝐼𝐼𝐼 − 𝜎𝐼𝐼

representing conjugate pairs of 𝜎𝑁 , 𝜎𝑆 values, which satisfy these equations, as a


point in the stress plane having 𝜎𝑁 as absicca and 𝜎𝑆 as ordinate.
Of the first equation, we note that

𝜎𝑁 − 𝜎𝐼𝐼 𝜎𝑁 − 𝜎𝐼𝐼𝐼 + 𝜎𝑆2 ≥ 0

08/07/2017 Dr. Luis Mosquera L. 109


For the case where the equality sign holds, this equation may be rewritten , to read

2 2
1 1
𝜎𝑁 − 𝜎𝐼𝐼 + 𝜎𝐼𝐼𝐼 + 𝜎𝑆2 = 𝜎 − 𝜎𝐼𝐼𝐼
2 2 𝐼𝐼

Which is the equation of the circle C1. The pairs of values 𝜎𝑁 and 𝜎𝑆 which satisfy
the inequality result in stress points having coordinates exterior to circle C1.
Following the same general procedure, we extract the equations of the circles C2
and C3 . The tree circles are called Mohr’s circles for stress.

08/07/2017 Dr. Luis Mosquera L. 110


All possible pairs values of 𝜎𝑁 and 𝜎𝑆 lie on these circles or between the areas
enclosed by them. In addition, it is clear from the Mohr’s circles diagram that the
maximum shear stress value at P is the radius of circle C2 .

In order to relate a typical stress point having coordinates 𝜎𝑁 and 𝜎𝑆 in


the stress plane:

08/07/2017 Dr. Luis Mosquera L. 111


Example:

The state of stress at point P is given in Mpa with respect to axes Px1x2x3 by the
matrix

25 0 0
𝑡𝑖𝑗 = 0 −30 −60
0 −60 5

(a) Determine the stress vector on the plane whose unit normal is

1
𝑛= 2𝑒1 + 𝑒2 + 2𝑒3
3

(b) Determine the normal stress component 𝜎𝑁 and 𝑠ℎ𝑒𝑎𝑟 𝑐𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡 𝜎𝑆 on the
same plane.

(c) Verify the results of part (b) by the Mohr’s circle construction

08/07/2017 Dr. Luis Mosquera L. 112


Solution

Verify that for the stress tensor tij given here the principal stress values are
σI = 50 Mpa, σII = 25 Mpa and σIII = -75 Mpa. Also, the transformation matrix from
axes Px1x2x3 to 𝑃𝑥1∗ 𝑥2∗ 𝑥3∗ is

0 −3/5 4/5
𝑎𝑖𝑗 = 1 0 0
0 4/5 3/5

So that the components of 𝑛 are given relative to the principal axes by

2
3
𝑛1∗ 0 −3/5 4/5 1/3
1
𝑛2∗ = 1 0 0 3 = 2/3
𝑛3∗ 0 4/5 3/5 2/3
2
3

08/07/2017 Dr. Luis Mosquera L. 113


Therefore, φ = cos-1 (1/3); β = θ = cos-1 (2/3), from which we may measure the
coordinates of the stress point q and confirm the values 𝜎𝑁 = −16,7 and 𝜎𝑆 =
52,7 in Mpa.

08/07/2017 Dr. Luis Mosquera L. 114


Plane Stress

When one, and only one principal stress is zero, we have a state of plane stress

08/07/2017 Dr. Luis Mosquera L. 115


If the principal stresses are not ordered and the direction of the zero principal stress
is arbitrarily chosen as x3 , the stress matrix takes the form

𝑡11 𝑡12 0
𝑡𝑖𝑗 = 𝑡12 𝑡22 0
0 0 0

Or, with respect to principal axes, the form


𝜎1 0 0
𝑡𝑖𝑗∗ = 0 𝜎2 0
0 0 0

The pictorial description of this plane


stress situation

08/07/2017 Dr. Luis Mosquera L. 116


This is sometimes represented by a single Mohr’s circle

The equation of the circle is

2 2
1 2
1 2
𝜎𝑁 − 𝑡11 + 𝑡22 + 𝜎𝑆 = 𝑡 − 𝑡22 + 𝑡12
2 2 11

08/07/2017 Dr. Luis Mosquera L. 117


Points A y B on the circle represent the stress states for area elements having unit
normals 𝑒1 and 𝑒2 respectively. For an element of area having a unit normal in an
arbitrary direction at point P, we must include the two dashed circles shown

08/07/2017 Dr. Luis Mosquera L. 118


With respect to axes 𝑂𝑥1′ 𝑥2′ 𝑥3′ rotated by the angle θ about the x3 axis

𝑡11 + 𝑡22 𝑡11 − 𝑡22



𝑡11 = + 𝑐𝑜𝑠2𝜃 + 𝑡12 𝑠𝑖𝑛2𝜃
2 2
𝑡11 + 𝑡22 𝑡11 − 𝑡22

𝑡22 = − 𝑐𝑜𝑠2𝜃 − 𝑡12 𝑠𝑖𝑛2𝜃
2 2
𝑡11 − 𝑡22

𝑡12 =− 𝑠𝑖𝑛2𝜃 + 𝑡12 𝑐𝑜𝑠2𝜃
2
08/07/2017 Dr. Luis Mosquera L. 119
If the principal axes of stress are chosen for the primed directions

2
1 1
𝜎1 , 𝜎2 = 𝑡11 + 𝑡22 ± 𝜎𝑆2 = 𝑡 − 𝑡22 2
+ 𝑡12
2 2 11

Example:
A specimen is loaded with equal tensile and shear stresses. This case of plane stress
may be represented by the matrix

𝜎0 𝜎0 0
𝑡𝑖𝑗 = 𝜎0 𝜎0 0
0 0 0

Where σ0 is a constant stress. Determine the principal values and plot the Mohr’s circles.

08/07/2017 Dr. Luis Mosquera L. 120


Solution

σ1 = 2σ0 , σ2 = σ3 = 0, so that, in principal axes form,

2𝜎0 0 0
𝑡𝑖𝑗∗ = 0 0 0
0 0 0

08/07/2017 Dr. Luis Mosquera L. 121


Desviator and spherical stress states

The arithmetic mean of the normal stresses,

1 1
𝜎𝑀 = 𝑡11 + 𝑡22 + 𝑡33 = 𝑡𝑖𝑖
3 3

Is referred to as the mean normal stress. The state of stress having all three principal
stresses equal is called a spherical state of stress

𝜎𝑀 0 0
𝑡𝑖𝑗 = 0 𝜎𝑀 0
0 0 𝜎𝑀

For which all directions are principal directions. The classical physical example is
the hydrostatic stress (-p0).

08/07/2017 Dr. Luis Mosquera L. 122


Every state of stress tij may be decomposed into a spherical portion and a portion Sij
known as the deviator stress

1
𝑡𝑖𝑗 = 𝑆𝑖𝑗 + 𝛿𝑖𝑗 𝜎𝑀 = 𝑆𝑖𝑗 + 𝛿𝑖𝑗 𝑡𝑘𝑘
3

We notice immediately that the first invariant of the deviator stress is

1
𝑆𝑖𝑖 = 𝑡𝑖𝑖 − 𝛿𝑖𝑖 𝑡𝑘𝑘 = 0
3
The characteristic equation for the deviator stress is

3
1
𝑆 − 𝑆𝑖𝑗 𝑆𝑗𝑖 − 𝜀𝑖𝑗𝑘 𝑆1𝑖 𝑆2𝑗 𝑆3𝑘
2

A principal direction of tij is also a principal direction of Sij . The principal values of Sij

𝑆𝑞 = 𝜎𝑞 − 𝜎𝑀
08/07/2017 Dr. Luis Mosquera L. 123
Example: Descompose the stress tensor T into its deviator and spherical portions and
determine the principal stress values of the deviator portion.

57 0 24
𝑡𝑖𝑗 = 0 50 0 MPa
24 0 43

08/07/2017 Dr. Luis Mosquera L. 124


Solution
1
𝜎𝑀 = 57 + 50 + 43 = 50 𝑀𝑃𝑎
3

Thus, decomposition leads to the matrix sum

57 0 24 7 0 24 50 0 0
0 50 0 = 0 0 0 + 0 50 0 𝑀𝑃𝑎
24 0 43 24 0 −7 0 0 50

Principal stress values of the deviator portion result from

7−𝑆 0 24
0 −𝑆 0 =0
24 0 −7 − 𝑆

Which yield SI = 25 Mpa, SII = 0 and SIII = -25 MPa


08/07/2017 Dr. Luis Mosquera L. 125
Problems

(10) The stress tensor at P relative to axes Px1x2x3 has components in MPa given by
the matrix representation

𝑡11 2 1
𝑡𝑖𝑗 = 2 0 2
1 2 0

Where t11 is unspecified. Determine a direction 𝑛 at P for which the plane


perpendicular to 𝑛 will be stress-free, that is, for which 𝑡 𝑛 = 0 on that plane.
What is the required value of t11 for this condition?

08/07/2017 Dr. Luis Mosquera L. 126


(11) Show that the sum of squares of the magnitudes of the stress vectors on the
coordinate planes is independent of the orientation of the coordinate axes, that is,
show that the sum

𝑒 𝑒 𝑒 𝑒 𝑒 𝑒
𝑡𝑖 1 𝑡𝑖 1 + 𝑡𝑖 2 𝑡𝑖 2 + 𝑡𝑖 3 𝑡𝑖 3

Is an invariant

08/07/2017 Dr. Luis Mosquera L. 127


(12) With respect to axes Ox1x2x3 the stress state is given in terms of the
coordinates by the matrix

𝑥1 𝑥2 𝑥22 0
𝑡𝑖𝑗 = 𝑥22 𝑥2 𝑥3 𝑥32
0 𝑥32 𝑥3 𝑥1

Determine

(a) The body force components as functions of the coordinates if the equilibrium
equations are to be satisfied everywhere, and
(b) The stress vector at point P (1,2,3) on the plane whose outward unit normal
makes equal angles with the positive coordinate axes.

08/07/2017 Dr. Luis Mosquera L. 128


(13) The stress components in a circular cylinder of length L and radius r are given by

𝐴𝑥2 + 𝐵𝑥3 𝐶𝑥3 −𝐶𝑥2


𝑡𝑖𝑗 = 𝐶𝑥3 0 0
−𝐶2 0 0

(a) Verify that in the absence of body forces the equilibrium equations are satisfied
(b) Show that the stress vector vanishes at all points on the curved surface of the
cylinder

08/07/2017 Dr. Luis Mosquera L. 129


(14) Sketch the Mohr’s circles for the various stress states shown on the cube which
is oriented along the coordinate axes

08/07/2017 Dr. Luis Mosquera L. 130


Kinematics of deformation and motion
Particles, configurations, deformations and motion

In continuum mechanics we consider material bodies in the form of solids, liquids


and gases. Whereas a particle of classical mechanics has an assigned mass, a
continuum particle is essentially a material point for which a density is defined.

x = k(X)

In which the vector function k assigns the position x relative to some origin for each
particle X of the body. Also,
X= k-1 (x)

identifies the particle X located at position x.

08/07/2017 Dr. Luis Mosquera L. 131


A arbitrary displacement will usually include both a rigid-body displacement and a
deformation which results in a change in size, or shape, or possibly both. A motion
of body B is a continuous time sequence of displacements that carries the set of
particles X into various configurations in a stationary space:

x = k(X,t)
Which gives the position x for each particle X for all times t.

We assume the motion function invertible and differentiable, so that

X = k-1(x,t)

Which identifies the particle X located at position x at time t.

08/07/2017 Dr. Luis Mosquera L. 132


Material and Spatial Coordinates
Consider that the position vector X of particle X relative to the axes OX1X2X3 is given by

𝑋 = ∅ 𝑋 = 𝑋𝐴 𝐼𝐴

XA are the material coordinates of the particle X

With respect to the usual cartesian axes Ox1x2x3 the current position vector is

𝒙 = 𝑘 ∅−1 𝑋 , 𝑡 = 𝑥𝑖 𝑒𝑖

xi are called the spatial coordinates of the particle

The material coordinates are used in conjuction with the reference configuration
only, and the spatial coordinates serve for all other configurations.

08/07/2017 Dr. Luis Mosquera L. 133


𝒙 = 𝑘 ∅−1 𝑋 , 𝑡 = 𝑥𝑖 𝑒𝑖

∅−𝟏 𝑿 maps the particle at X in the reference configuration onto the point x
in the current configuration at time t as indicated in the figure.

08/07/2017 Dr. Luis Mosquera L. 134


x = x(X,t)

x = x(X,0)= X at time t =0 (the reference configuration)

The initial spatial coordinates being identical in value with the material coordinates

IF we focus attention on a specific particle XP, then

𝑥 𝑃 = 𝑥(𝑋 𝑃 , 𝑡)
xP describe the path, or trajectory of that particle as a function of time. The velocity
of the particle along its path is defined as

𝜕𝑥
𝑣𝑃 = 𝑥𝑃 =
𝜕𝑡 𝑋=𝑋 𝑃

08/07/2017 Dr. Luis Mosquera L. 135


Of course, the individual particles of a body cannot execute arbitrary motions
independent of one another. No two particles can occupy the same location in space
at a given time. For that reason, the function X must be single-valued and
continuous, and must possess continuous derivatives with respect to space and time.

For know where the particle , now at x, was located in the reference configuration,
we require the inverse function x-1

𝑋 = 𝑥 −1 (𝑥, 𝑡)

The condition that guarantees the existence of such an inverse function is

𝜕𝑥𝑖
𝐽= ≠0
𝜕𝑋𝐴

08/07/2017 Dr. Luis Mosquera L. 136


Example: Let the motion of a body be given by

𝑥1 = 𝑋1 + 𝑡 2 𝑋2
𝑥2 = 𝑋2 + 𝑡 2 𝑋1
𝑥3 = 𝑋3

Determine
(a) The path of the particle originally at X = (1,2,1)
(b) The velocity when t = 2 s

08/07/2017 Dr. Luis Mosquera L. 137


Solution

(a) For X = (1,2,1) the motion equations are

𝑥1 = 1 + 2𝑡 2
𝑥2 = 2 + 𝑡 2
𝑥3 = 1

Which upon elimination of the variable t gives x1 – 2x2 = -3 as well as x3 =1 so


that the particle moves on a straight line path in the plane x3 = 1.

08/07/2017 Dr. Luis Mosquera L. 138


Lagrangian and Eulerian Descriptions

If a physical property of the body B such as its density ρ, or a kinematic property of its
motion such as the velocity v, is expressed in terms of the material coordinates X,
and the time t, we say that property is given by the material description.

When the referential configuration is taken as the actual configuration at time t = 0,


this description is usually called the Lagrangian description. Thus, the equations

𝜌 = 𝜌(𝑋, 𝑡)

𝑣 = 𝑣(𝑋, 𝑡)

chronicle a time history of these properties for each particle of the body.

08/07/2017 Dr. Luis Mosquera L. 139


If the properties ρ and v are given as functions of the spatial coordinates x
and time t, we say that those properties are expressed by a spatial
description or by the Eulerian description.

In the material description attention is focused on what is


happening to the individual particles during the motion,
whereas in the spatial description the emphasis is directed to
the events taking place at specific points in space.

08/07/2017 Dr. Luis Mosquera L. 140


Example: Let the motion equations be given in component form by the Lagrangian
description

𝑥1 = 𝑋1 𝑒 𝑡 + 𝑋3 (𝑒 𝑡 − 1)

𝑥2 = 𝑋2 + 𝑋3 (𝑒 𝑡 − 𝑒 −𝑡 )

𝑥3 = 𝑋3

Determine la Eulerian description of this motion

08/07/2017 Dr. Luis Mosquera L. 141


Solution

Notice that for the given motion 𝑥1 = 𝑋1 , 𝑥2 = 𝑋2 , 𝑥3 = 𝑋3 at t = 0, so that the


initial configuration has been taken as the reference configuration.

𝑋1 = 𝑥1 𝑒 −𝑡 + 𝑥3 (𝑒 −𝑡 − 1)

𝑋2 = 𝑥2 + 𝑥3 (𝑒 −𝑡 − 𝑒 𝑡 )

𝑋3 = 𝑥3

Eulerian description

08/07/2017 Dr. Luis Mosquera L. 142


The Displacement Field

The typical particle of body P undergoes a displacement u = x – X in the transition


from the reference configuration to the current configuration.

08/07/2017 Dr. Luis Mosquera L. 143


Because this relationship holds for all particles it is often useful to analyze deformation
or motion in terms of the displacement field of the body.

𝑢 = 𝑢𝑖 𝑒𝑖 = 𝑢𝐴 𝐼𝐴

The material form: we are describing the displacement that will occur to the
particle that stars at X

𝑢 𝑋, 𝑡 = 𝑥 𝑋, 𝑡 − 𝑋

The spatial form: present the displacement that the particle now at x has undergone

𝑢 𝑥, 𝑡 = 𝑥 − 𝑋(𝑥, 𝑡)

08/07/2017 Dr. Luis Mosquera L. 144


We may take the time rate of change of displacement as an alternative definition
for velocity

𝑑𝑢 𝑑(𝑥 − 𝑋) 𝑑𝑥
= = =𝑣
𝑑𝑡 𝑑𝑡 𝑑𝑡

Example: Obtain the displacement field for the motion

𝑥1 = 𝑋1 𝑒 𝑡 + 𝑋3 (𝑒 𝑡 − 1)

𝑥2 = 𝑋2 + 𝑋3 (𝑒 𝑡 − 𝑒 −𝑡 )

𝑥3 = 𝑋3

08/07/2017 Dr. Luis Mosquera L. 145


Solution

We may compute the displacement field in material form directly as

𝑢1 = 𝑥1 − 𝑋1 = (𝑋1 + 𝑋3 )(𝑒 𝑡 − 1)

𝑢2 = 𝑥2 − 𝑋2 = 𝑋3 (𝑒 𝑡 − 𝑒 −𝑡 )

𝑢3 = 𝑥3 − 𝑋3 = 0

The spatial description

𝑢1 = (𝑥1 + 𝑥3 )(1 − 𝑒 −𝑡 )

𝑢2 = 𝑥3 (𝑒 𝑡 − 𝑒 −𝑡 )

𝑢3 = 0

08/07/2017 Dr. Luis Mosquera L. 146


The Material Derivative

Consider any physical or kinematic property of a continuum body. It may be a


scalar, vector or tensor property; we represent it by Pij…

In the material description 𝑃𝑖𝑗 … = 𝑃𝑖𝑗 … (𝑋, 𝑡)

Or in the spatial description 𝑃𝑖𝑗 … = 𝑃𝑖𝑗 … (𝑥, 𝑡)

The material derivative can be thought of as the rate at which Pij… changes when
measured by an observer attached to, and traveling with, the particle, or group of
particles.

08/07/2017 Dr. Luis Mosquera L. 147


When Pij… is given in the material description, the material derivative is simply the
partial derivative with respect to time,

𝑑 𝜕
𝑃𝑖𝑗 … (𝑋, 𝑡) = 𝑃 (𝑋, 𝑡)
𝑑𝑡 𝜕𝑡 𝑖𝑗 …

If, Pij… is given in the spatial form, we recognize that the specific collection of particles
of interest will be changing position in space, and we obtain:

𝑑 𝜕 𝜕 𝑑𝑥𝑘
𝑃𝑖𝑗 … (𝑥, 𝑡) = 𝑃𝑖𝑗 … (𝑥, 𝑡) + 𝑃𝑖𝑗 … (𝑥, 𝑡)
𝑑𝑡 𝜕𝑡 𝜕𝑥𝑘 𝑑𝑡

In this equation, the term 𝜕


𝑃 (𝑥, 𝑡)
𝜕𝑡 𝑖𝑗 …

gives the change occurring in the property at position x, known as the local rate
of change.

08/07/2017 Dr. Luis Mosquera L. 148


The term 𝜕 𝑑𝑥𝑘
𝑃𝑖𝑗 … (𝑥, 𝑡) results from the particles changing position in
𝜕𝑥𝑘 𝑑𝑡

space and is referred to as the convective rate of change.


𝑑𝑥 𝑑𝑥𝑘
As 𝑣= 𝑜𝑟 𝑣𝑘 =
𝑑𝑡 𝑑𝑡

Then, 𝑑 𝜕 𝜕
𝑃𝑖𝑗 … (𝑥, 𝑡) = 𝑃𝑖𝑗 … (𝑥, 𝑡) + 𝑃 (𝑥, 𝑡) 𝑣𝑘
𝑑𝑡 𝜕𝑡 𝜕𝑥𝑘 𝑖𝑗 …

From which we deduce the material derivative operator for properties expressed in
the spatial description
𝑑 𝜕 𝜕
= + 𝑣𝑘
𝑑𝑡 𝜕𝑡 𝜕𝑥𝑘
𝑑 𝜕
= + 𝑣. ∇
𝑑𝑡 𝜕𝑡

08/07/2017 Dr. Luis Mosquera L. 149


Example: Let a certain motion of a continuum be given by the component equations,

𝑥1 = 𝑋1 𝑒 −𝑡

𝑥2 = 𝑋2 𝑒 𝑡

𝑥3 = 𝑋3 + 𝑋2 (𝑒 −𝑡 − 1)

and let the temperature field of the body be given by the spatial description.

𝜃 = 𝑒 −𝑡 (𝑥1 − 2𝑥2 + 3𝑥3 )

Determine the velocity field in spatial form, and using that, compute the material
derivative dθ/dt of the temperature field.

08/07/2017 Dr. Luis Mosquera L. 150


Solution

Note that the initial configuration serves as the reference configuration.


The velocity components in material form are
𝑣1 = −𝑋1 𝑒 −𝑡

𝑣2 = 𝑋2 𝑒 𝑡

𝑣3 = −𝑋2 𝑒 −𝑡

Also, the motion equations can be inverted directly to give,

𝑋1 = 𝑥1 𝑒 𝑡

𝑋2 = 𝑥2 𝑒 −𝑡

𝑋3 = 𝑥3 − 𝑥2 (𝑒 −2𝑡 − 𝑒 −𝑡 )

08/07/2017 Dr. Luis Mosquera L. 151


Which yields the spatial components for the velocity,

𝑣1 = −𝑥1

𝑣2 = 𝑥2

𝑣3 = −𝑥2 𝑒 −2𝑡

Therefore, dθ/dt in spatial form

𝑑𝜃
= −𝑒 −𝑡 𝑥1 − 2𝑥2 + 3𝑥3 − 𝑥1 𝑒 −𝑡 − 2𝑥2 𝑒 −𝑡 − 3𝑥3 𝑒 −𝑡
𝑑𝑡

Which converted to its material form result in

𝑑𝜃
= −2𝑋1 𝑒 −2𝑡 − 3𝑋2 (2𝑒 −2𝑡 − 𝑒 −𝑡 ) − 3𝑋3 𝑒 −𝑡
𝑑𝑡

08/07/2017 Dr. Luis Mosquera L. 152


Strain Tensors
The deformation gradient tensor
In deformation analysis we confine our attention to two stationary configurations and
disregard any consideration for the particular sequence by which the final deformed
configuration is reached from the initial undeformed configuration.

08/07/2017 Dr. Luis Mosquera L. 153


Accordingly, the mapping function is not dependent upon time as a variable, so that

𝑥𝑖 = 𝑋𝑖 (𝑋)

Consider 𝑑𝑋 = 𝑑𝑋𝐴 𝐼𝐴
2
𝑑𝑋 = 𝑑𝑋. 𝑑𝑋 = 𝑑𝑋𝐴 𝑑𝑋𝐴

Under the displacement field, the particles originally at P and Q move to the
positions p and q, respectively, in the deformed configuration. Now

𝑑𝑥 = 𝑑𝑥𝑖 𝑒𝑖
2
𝑑𝑥 = 𝑑𝑥. 𝑑𝑥 = 𝑑𝑥𝑖 𝑑𝑥𝑖

08/07/2017 Dr. Luis Mosquera L. 154


We assume the mapping function Xi is continuous so that

𝜕𝑋𝑖
𝑑𝑥𝑖 = 𝑑𝑋 = 𝑥𝑖,𝐴 𝑑𝑋𝐴
𝜕𝑋𝐴 𝐴
where
𝑥𝑖,𝐴 ≡ 𝐹𝑖𝐴

Is called the deformation gradient tensor. If F depend od X the deformation is


termed inhomogeneous, otherwise, if F is independent of X, the deformation is
called homogeneous.

In symbolic notation
𝑑𝑥 = 𝐹. 𝑑𝑋 𝑜𝑟 𝑑𝑥 = 𝐹𝑑𝑋

08/07/2017 Dr. Luis Mosquera L. 155


F is invertible so that the inverse F-1 exist such that

𝑑𝑋 = 𝐹 −1 𝑑𝑥

Let us consider
2 2
𝑑𝑥 − 𝑑𝑋 = 𝑑𝑥𝑖 𝑑𝑥𝑖 − 𝑑𝑋𝐴 𝑑𝑋𝐴

Which and the substitution property of the Kronecker delta,

2 2
𝑑𝑥 − 𝑑𝑋 = 𝑥𝑖,𝐴 𝑑𝑋𝐴 𝑥𝑖,𝐵 𝑑𝑋𝐵 − 𝛿𝐴𝐵 𝑑𝑋𝐴 𝑑𝑋𝐵
= 𝑥𝑖,𝐴 𝑥𝑖,𝐵 − 𝛿𝐴𝐵 𝑑𝑋𝐴 𝑑𝑋𝐵
= 𝐶𝐴𝐵 − 𝛿𝐴𝐵 𝑑𝑋𝐴 𝑑𝑋𝐵

08/07/2017 Dr. Luis Mosquera L. 156


Where the symmetric tensor

𝐶𝐴𝐵 = 𝑥𝑖,𝐴 𝑥𝑖,𝐵 𝑜𝑟 𝐶 = 𝐹 𝑇 𝐹

Is called the Green’s deformation tensor. From this define the Lagrangian finite
strain tensor EAB as

2𝐸𝐴𝐵 = 𝐶𝐴𝐵 − 𝛿𝐴𝐵 𝑜𝑟 2𝐸 = 𝐶 − 𝐼

Finally we can write,

2 2
𝑑𝑥 − 𝑑𝑋 = 2𝐸𝐴𝐵 𝑑𝑋𝐴 𝑑𝑋𝐵 = 𝑑𝑋. 2𝐸. 𝑑𝑋

08/07/2017 Dr. Luis Mosquera L. 157


In a similar way, in terms of the spatial variables

2 2
𝑑𝑥 − 𝑑𝑋 = 𝛿𝑖𝑗 𝑑𝑥𝑖 𝑑𝑥𝑗 − 𝑋𝐴,𝑖 𝑑𝑥𝑖 𝑋𝐴,𝑗 𝑑𝑥𝑗
= 𝛿𝑖𝑗 − 𝑋𝐴,𝑖 𝑋𝐴,𝑗 𝑑𝑥𝑖 𝑑𝑥𝑗
= 𝛿𝑖𝑗 − 𝑐𝑖𝑗 𝑑𝑥𝑖 𝑑𝑥𝑗

Where the symmetric tensor

𝑐𝑖𝑗 = 𝑋𝐴,𝑖 𝑋𝐴,𝑗 𝑜𝑟 𝑐 = (𝐹 −1 )𝑇 . 𝐹 −1

Is called the Cauchy deformation tensor

Also is defined the Eulerian finite strain tensor e as

2𝑒𝑖𝑗 = 𝛿𝑖𝑗 − 𝑐𝑖𝑗 𝑜𝑟 2𝑒 = (𝐼 − 𝑐)

2 2
𝑑𝑥 − 𝑑𝑋 = 2𝑒𝑖𝑗 𝑑𝑥𝑖 𝑑𝑥𝑗 = 𝑑𝑥. 2𝑒. 𝑑𝑥

08/07/2017 Dr. Luis Mosquera L. 158


For any two arbitrary differential vectors dX(1) and dX(2) which deform into dx(1) and
dx(2), respectively, we have

𝑑𝑥 (1) . 𝑑𝑥 (2) = 𝐹. 𝑑𝑋 (1) . 𝐹. 𝑑𝑋 (2)


= 𝑑𝑋 (1) . 𝐹 𝑇 . 𝐹. 𝑑𝑋 (2)
= 𝑑𝑋 1 . 𝐼 + 2𝐸 . 𝑑𝑋 (2)

If E is identically zero, the lengths of all line elements are unchanged so that (dx)2 = (dX)2
and in view of the definition dx(1) . dx(2) = dx(1) dx(2) cosθ, the angle between any two
elements will also be unchanged. Thus in the absence of strain, only a rigid body
displacement can occur.

08/07/2017 Dr. Luis Mosquera L. 159


The Lagrangian and Eulerian finite strain tensors may also be developed in terms of
displacement gradients.

The material description is

𝑢𝑖 𝑋𝐴 = 𝑥𝑖 𝑋𝐴 − 𝑋𝑖

2𝐸𝐴𝐵 = 𝑢𝐴,𝐵 + 𝑢𝐵,𝐴 + 𝑢𝑖,𝐴 𝑢𝑖,𝐵

And the spatial description is

𝑢𝐴 𝑥𝑖 = 𝑥𝐴 − 𝑋𝐴 (𝑥𝑖 )

2𝑒𝑖𝑗 = 𝑢𝑖,𝑗 + 𝑢𝑗 ,𝑖 − 𝑢𝐴,𝑖 𝑢𝐴,𝑗

08/07/2017 Dr. Luis Mosquera L. 160


Example: Let the simple shear deformation

𝑥1 = 𝑋1 ; 𝑥2 = 𝑋2 + 𝑘𝑋3 ; 𝑥3 = 𝑋3 + 𝑘𝑋2

Where k is a constant, be applied to the small cube of edge dimensions dL. Draw the
deformed shape of face ABGH of the cube and determine the difference (dx)2 – (dX)2
for the diagonals AG,BH and OG of the cube.

08/07/2017 Dr. Luis Mosquera L. 161


Solution

From the mapping equations directly, the origin O is seen to remain in place, and
the particles originally at points A,B,G and H are displaced to the points a(dL,O,O),
b(dL,dL,kdL), g(dL, (1+k)dL,(1+k)dL) and h(dL,kdL,dL), respectively, and the square
face ABGH becomes the diamond shaped parallelogram abgh.

08/07/2017 Dr. Luis Mosquera L. 162


1 0 0
𝑥𝑖,𝐴 ≡ 𝐹𝑖𝐴 𝐹𝑖𝐴 = 0 1 𝑘
0 𝑘 1

1 0 0
𝑇
𝐶=𝐹 𝐹 𝐶𝐴𝐵 = 0 1 + 𝑘2 2𝑘
0 2𝑘 1 + 𝑘2

0 0 0
2𝐸 = 𝐶 − 𝐼 2𝐸𝐴𝐵 = 0 𝑘2 2𝑘
0 2𝑘 𝑘2

08/07/2017 Dr. Luis Mosquera L. 163


2 2
𝑑𝑥 − 𝑑𝑋 = 𝑑𝑋. 2𝐸. 𝑑𝑋 so that for diagonal AG,

0 0 0 0
𝑑𝑥 2 − 𝑑𝑋 2 = 0 𝑑𝐿 𝑑𝐿 0 𝑘 2 2𝑘 𝑑𝐿
0 2𝑘 𝑘 2 𝑑𝐿
2 2
= 2(2𝑘 + 𝑘 )(𝑑𝐿)

for diagonal BH,


0 0 0 0
2 2
𝑑𝑥 − 𝑑𝑋 = 0 −𝑑𝐿 𝑑𝐿 0 𝑘 2 2𝑘 −𝑑𝐿
0 2𝑘 𝑘2 𝑑𝐿
2 2
= 2(−2𝑘 + 𝑘 )(𝑑𝐿)

and for diagonal OG,


0 0 0 𝑑𝐿
2 2
𝑑𝑥 − 𝑑𝑋 = 𝑑𝐿 𝑑𝐿 𝑑𝐿 0 𝑘 2 2𝑘 𝑑𝐿
0 2𝑘 𝑘 2 𝑑𝐿
= 2(2𝑘 + 𝑘 2 )(𝑑𝐿)2

08/07/2017 Dr. Luis Mosquera L. 164


Infinitesimal deformation theory

If the numerical values of all the components of the displacement and the displacement
gradient tensors are very small we may neglect the squares and products of these
quantities in comparison to the gradients themselves so that

2𝐸𝐴𝐵 = 𝑢𝐴,𝐵 + 𝑢𝐵,𝐴


and
2𝑒𝑖𝑗 = 𝑢𝑖,𝑗 + 𝑢𝑗 ,𝑖

Furthermore, to the same order of approximation,

𝜕𝑢𝑖 𝜕𝑢𝑖 𝜕𝑥𝑘 𝜕𝑢𝑖 𝜕𝑢𝑘 𝜕𝑢𝑖


= = + 𝛿𝑘𝐴 ≈ 𝛿𝑘𝐴
𝜕𝑋𝐴 𝜕𝑥𝑘 𝜕𝑋𝐴 𝜕𝑥𝑘 𝜕𝑋𝐴 𝜕𝑥𝑘

Therefore, to the first order of approximation for the case of small displacement
gradients, it is unimportant whether we differentiate the displacement
components with respect to the material or spatial coordinates.
08/07/2017 Dr. Luis Mosquera L. 165
Similarly, it can be shown that in the linear theory uA,B and uA,j are equivalent

𝐸𝐴𝐵 ≈ 𝑒𝑖𝑗 𝛿𝑖𝐴 𝛿𝑗𝐵

And it is customary to define a single infinitesimal strain tensor εij as

𝜕𝑢𝑖 𝜕𝑢𝑗 𝜕𝑢𝑖 𝜕𝑢𝑗


2𝜀𝑖𝑗 = 𝛿 + 𝛿 = + = 𝑢𝑖,𝑗 + 𝑢𝑗 ,𝑖
𝜕𝑋𝐴 𝐴𝑗 𝜕𝑋𝐵 𝐵𝑖 𝜕𝑥𝑗 𝜕𝑥𝑖

The matrix, relative to principal axes, for εij is

𝜀1 0 0 𝜀𝐼 0 0
𝜀𝑖𝑗∗ = 0 𝜀2 0 = 0 𝜀𝐼𝐼 0
0 0 𝜀3 0 0 𝜀𝐼𝐼𝐼

08/07/2017 Dr. Luis Mosquera L. 166


Together with the strain invariants

𝐼𝜀 = 𝜀𝑖𝑖 = 𝑡𝑟 𝜀 = 𝜀𝐼 + 𝜀𝐼𝐼 +𝜀𝐼𝐼𝐼


1
𝐼𝐼𝜀 = 𝜀𝑖𝑖 𝜀𝑗𝑗 − 𝜀𝑖𝑗 𝜀𝑖𝑗 = 𝜀𝐼 𝜀𝐼𝐼 + 𝜀𝐼𝐼 𝜀𝐼𝐼𝐼 + 𝜀𝐼𝐼𝐼 𝜀𝐼
2
𝐼𝐼𝐼𝜀 = 𝜖𝑖𝑗𝑘 𝜀1𝑖 𝜀2𝑗 𝜀3𝑘 = 𝜀𝐼 𝜀𝐼𝐼 𝜀𝐼𝐼𝐼

08/07/2017 Dr. Luis Mosquera L. 167


The components of ε, physical interpretations

Within the context of small deformation,

2 2
𝑑𝑥 − 𝑑𝑋 = 2𝜀𝑖𝑗 𝑑𝑋𝑖 𝑑𝑋𝑗 = 𝑑𝑋. 2𝜀. 𝑑𝑋

𝑑𝑥 − 𝑑𝑋 𝑑𝑥 + 𝑑𝑋 𝑑𝑋𝑖 𝑑𝑋𝑗
. = 2𝜀𝑖𝑗
𝑑𝑋 𝑑𝑋 𝑑𝑋 𝑑𝑋

But 𝑑𝑋𝑖 a unit vector in the direction of dX, so that


= 𝑁𝑖
𝑑𝑋
𝑑𝑥 − 𝑑𝑋
= 𝜀𝑖𝑗 𝑁𝑖 𝑁𝑗 = 𝑁. 𝜀. 𝑁
𝑑𝑋

𝑑𝑥 − 𝑑𝑋
Longitudinal strain 𝑒𝑁 =
𝑑𝑋

08/07/2017 Dr. Luis Mosquera L. 168


If N is taken in the X1 direction so that 𝑁 = 𝐼 1 , then

𝑒 𝐼1 = 𝐼1 . 𝜀. 𝐼1 = 𝜀11

Likewise, for N = I2 or N = I3 the normal strains are found to be ε22 and ε33 .The
diagonal elements of the infinitesimal strain tensor represent normal strains in the
coordinate directions.

If consider the differential vectors dX(1) and dX(2) at position P which are deformed
into vectors dx(1) and dx(2), may be written,

𝑑𝑥 (1) . 𝑑𝑥 (2) = 𝑑𝑋 (1) . 𝑑𝑋 (2) + 𝑑𝑋 1


. 2𝜀. 𝑑𝑋 (2)

If dX(1) and dX(2) are perpendicular to one another,

𝑑𝑥 (1) . 𝑑𝑥 (2) = 𝑑𝑥 (1) 𝑑𝑥 (2) 𝑐𝑜𝑠𝜃 = 𝑑𝑋 1


. 2𝜀. 𝑑𝑋 (2)
Where θ is the angle between the deformed vectors

08/07/2017 Dr. Luis Mosquera L. 169


If: 𝜋
𝜃 = −𝛾
2
𝜋
𝑐𝑜𝑠𝜃 = 𝑐𝑜𝑠 − 𝛾 = 𝑠𝑖𝑛𝛾 ≈ 𝛾
2

Since ϒ is very small for infinitesimal deformations. Therefore,

𝑑𝑋 1 𝑑𝑋 (2)
𝛾 ≈ 𝑐𝑜𝑠𝜃 = 1
. 2𝜀. (2) ≈ 𝑁1 . 2𝜀. 𝑁2
𝑑𝑥 𝑑𝑥

08/07/2017 Dr. Luis Mosquera L. 170


Here, if we take 𝑁(1) = 𝐼 1 and 𝑁 (2)= 𝐼 2 and designate the angle ϒ as ϒ12, we obtain

𝜀11 𝜀12 𝜀13 0


𝛾12 = 2 1 0 0 𝜀21 𝜀22 𝜀23 1 = 2𝜀12
𝜀31 𝜀32 𝜀33 0

So that 𝛾𝑖𝑗 = 2𝜀𝑖𝑗 𝑖≠𝑗

The engineering shear strain components ϒij represent the changes in the original right
angles between the coordinate axes in the undeformed configuration.

08/07/2017 Dr. Luis Mosquera L. 171


In engineering texts, the infinitesimal strain tensor is frequently written in matrix
form as

1 1
𝜀11 𝛾 𝛾
2 12 2 13
1 1
𝜀𝑖𝑗 = 𝛾 𝜀22 𝜀 𝛾23
2 12 2
1 1
𝛾 𝛾 𝜀33
2 13 2 23

If 𝑁(1) and 𝑁 (2) are chosen in principal strain directions, then

𝛾 = 𝑁1 . 2𝜀 ∗ . 𝑁2 = 0

From which we may generalize to conclude that principal strain directions


remain orthogonal under infinitesimal deformation.

08/07/2017 Dr. Luis Mosquera L. 172


Therefore, a small rectangular parallelpied of undeformed edge dimensions dX(1)
, dX(2) and dX(3) taken in the principal strain directions will be deformed into
another rectangular parallelpiped having edge lengths

𝑑𝑥 (𝑖) = (1 + 𝜀 𝑖 )𝑑𝑋 (𝑖)

Where ε(i) are the normal strains in principal directions.

08/07/2017 Dr. Luis Mosquera L. 173


The change in volume per unit original volume of the parallelpiped is

∆𝑉 1 + 𝜀(1) 𝑑𝑋 (1) 1 + 𝜀(2) 𝑑𝑋 (2) 1 + 𝜀(3) 𝑑𝑋 (3) − 𝑑𝑋 (1) 𝑑𝑋 (2) 𝑑𝑋 (3)


=
𝑉 𝑑𝑋 (1) 𝑑𝑋 (2) 𝑑𝑋 (3)
≈ 𝜀(1) + 𝜀(2) + 𝜀(3)

The ratio ΔV/V being the first invariant of ε, is called the cubical dilation

08/07/2017 Dr. Luis Mosquera L. 174


Example

A delta rosette has the shape of an equilateral triangle, and records longitudinal
strains in the directions x1, x´1 and x´´1 shown in the sketch. If the measured strains in
these directions are ε11 = -3x10-4, ε´11 = 4x10-4 and ε´´11 = 2x10-4 , determine ε22 , ϒ12
and ε´22 . Show that ε11 + ε22 = ε´11 + ε´22 as the first strain invariant requires.

08/07/2017 Dr. Luis Mosquera L. 175


Solution


𝜀11 + 𝜀22 𝜀11 − 𝜀22 𝛾12
𝜀11 = + 𝑐𝑜𝑠2𝜃 + 𝑠𝑖𝑛2𝜃
2 2 2

𝜀11 + 𝜀22 𝜀11 − 𝜀22 𝛾12
𝜀22 = − 𝑐𝑜𝑠2𝜃 − 𝑠𝑖𝑛2𝜃
2 2 2

𝛾12 = − 𝜀11 − 𝜀22 𝑠𝑖𝑛2𝜃 + 𝛾12 𝐶𝑜𝑠2𝜃

Thus, for θ = 60° and θ = 120° respectively, we have

−3 + 𝜀22 −3 − 𝜀22 1 𝛾12 3


4= + (− ) +
2 2 2 2 2
−3 + 𝜀22 −3 − 𝜀22 1 𝛾12 3
2= + (− ) −
2 2 2 2 2

08/07/2017 Dr. Luis Mosquera L. 176


Where we determine ε22 = 5 and ϒ12 = 4/31/2. Next, using θ = 150° we determine ε’22


−3 + 5 −3 − 5 2 3
𝜀22 = + + − = −2
2 2 3 2

08/07/2017 Dr. Luis Mosquera L. 177


Consider once more two particles P and Q in the undeformed configuration, and
are now at positions p and q, in the deformed configuration. In general, an
arbitrary displacement will include both deformation (strain) and rigid boy
displacements. To separate these we consider the vector du,

𝜕𝑢𝑖
𝑑𝑢𝑖 = 𝑑𝑋𝑗
𝜕𝑋𝑗
𝑃

𝑑𝑢𝑖 𝑑𝑢𝑖 𝑑𝑋𝑗 𝑑𝑢𝑖


= = 𝑁
𝑑𝑋 𝑑𝑋𝑗 𝑑𝑋 𝑑𝑋𝑗 𝑗

Where Nj is the unit vector in the direction from P toward Q. By decomposing the
displacement gradient into its symmetric and skew-symmetric parts,

1 𝜕𝑢𝑖 𝜕𝑢𝑗 1 𝜕𝑢𝑖 𝜕𝑢𝑗


𝑑𝑢𝑖 = + + − 𝑑𝑋𝑗
2 𝜕𝑋𝑗 𝜕𝑋𝑖 2 𝜕𝑋𝑗 𝜕𝑋𝑖
= 𝜀𝑖𝑗 + 𝑤𝑖𝑗 𝑑𝑋𝑗
08/07/2017 Dr. Luis Mosquera L. 178
If εij happens to be identically zero, there is no strain, and the displacement is a
rigid body displacement. For this case is define the rotation vector

1
𝑤𝑖 = 𝜀𝑖𝑗𝑘 𝑤𝑘𝑗
2

𝑤𝑖𝑗 = 𝜀𝑘𝑗𝑖 𝑤𝑘

Therefore, with εij = 0

𝑑𝑢𝑖 = 𝜀𝑘𝑗𝑙 𝑤𝑘 𝑑𝑋𝑗 = 𝜀𝑖𝑘𝑗 𝑤𝑘 𝑑𝑋𝑗

or
𝑑𝑢 = 𝑤 × 𝑑𝑋

So that the relative differential displacement is seen to be the result of a rigid


body rotation about the axis of the rotation vector w. On the other hand, si wij =0,
the relative displacement will be the result of pure strain.

08/07/2017 Dr. Luis Mosquera L. 179


P15.- The motion of a continuous medium is specified by the component equations

1 1
𝑥1 = 𝑋1 + 𝑋2 𝑒 + 𝑋1 − 𝑋2 𝑒 −𝑡
𝑡
2 2
1 1
𝑥2 = 𝑋1 + 𝑋2 𝑒 − 𝑋1 − 𝑋2 𝑒 −𝑡
𝑡
2 2
𝑥3 = 𝑋3

a) Show that the Jacobian determinant J does not vanish, and solve for the inverse
equations X=X(x,t).
b) Calculate the velocity and acceleration components in terms of the material
coordinates.
c) Using the inverse equations developed in part (a), express the velocity and
acceleration components in terms of spatial coordinates.

08/07/2017 Dr. Luis Mosquera L. 180


P16.- A velocity field is given in Lagrangian form by

𝑣1 = 2𝑡 + 𝑋1

𝑣2 = 𝑋2 𝑒 𝑡

𝑣3 = 𝑋3 − 𝑡

Integrate these equations to obtain x = x(X,t) with x = X at t = 0, and using that result
compute the velocity and acceleration components in the Eulerian (spatial) form.

08/07/2017 Dr. Luis Mosquera L. 181


P17.- For the deformation field given by

𝑥1 = 𝑋1 + 𝛼𝑋2

𝑥2 = 𝑋1 − 𝛼𝑋2

𝑥3 = 𝑋3

Where α is a constant, determine the matrix form of the tensors E and e, and show

that the circle of particles 𝑋12 + 𝑋22 = 1 deforms into the circle 𝑥12 + 𝑥22 = 1 + 𝛼 2

08/07/2017 Dr. Luis Mosquera L. 182


Fundamental Laws and Equations
Compatibility Equations

If we consider the six independent strain-displacement relations,

𝜕𝑢𝑖 𝜕𝑢𝑗
+ = 2𝜀𝑖𝑗
𝜕𝑥𝑗 𝜕𝑥𝑖

for determining the ui, then the system is over-determined, and we cannot in general
find three single-valued functions ui = ui (xj) satisfying the six partial differential
equations. Therefore, some restrictive conditions must be imposed upon the strain
components if the equations above are to be satisfied by a single-valued displacement
field.

The strain compatibility equations

𝜀𝑖𝑗 ,𝑘𝑚 + 𝜀𝑘𝑚 ,𝑖𝑗 − 𝜀𝑖𝑘 ,𝑗𝑚 − 𝜀𝑗𝑚 ,𝑖𝑘 = 0

08/07/2017 Dr. Luis Mosquera L. 183


There are 81 equations in all but only six of these are distinct

𝜀11,23 + 𝜀23,11 − 𝜀12,13 − 𝜀13,12 = 0


𝜀22,31 + 𝜀31,22 − 𝜀23,21 − 𝜀21,23 = 0
𝜀33,12 + 𝜀12,33 − 𝜀31,32 − 𝜀32,31 = 0
2𝜀12,12 − 𝜀11,22 − 𝜀22,11 = 0
2𝜀23,23 − 𝜀22,33 − 𝜀33,22 = 0
2𝜀31,31 − 𝜀33,11 − 𝜀11,33 = 0

08/07/2017 Dr. Luis Mosquera L. 184


For plane strain in the x1-x2 plane, the six equations reduce to a single equation,

𝜀11,22 + 𝜀22,11 = 2𝜀12,12

For demonstrated the necessity and sufficiency of the compatibility equations,


given the strain εij with continuously differentiable displacements

𝑐𝑢𝑟𝑙 𝑐𝑢𝑟𝑙 𝜀 = 0
By the definition of linear strain and the continuity of the displacements, we see

𝜖𝑖𝑗𝑘 𝜖𝑙𝑚𝑛 𝜀𝑗𝑚 ,𝑘𝑛 = 𝜖𝑖𝑗𝑘 𝜖𝑙𝑚𝑛 𝑢𝑗 ,𝑚𝑘𝑛 + 𝑢𝑚 ,𝑗𝑘𝑛 = 0

Where the first displacement term, symmetric in mn, cancel with skew-simmetric
ϵlmn, and the second displacement term, simmetric in jk, cancels with skew-term ϵijk

08/07/2017 Dr. Luis Mosquera L. 185


To demonstrate sufficiency it must be shown that starting from curl curl ε = 0 that
the linear strain eij can be constructed.

𝐴 = 𝑐𝑢𝑟𝑙 𝜀

Compatibility gives
𝑐𝑢𝑟𝑙 𝐴 = 0

Since ε is symmetric, 𝑡𝑟 𝐴 = 0

Also, there exists a skew-symmetric tensor w such that 𝐴 = −𝑐𝑢𝑟𝑙 𝑤

Therefore, 𝑐𝑢𝑟𝑙 𝜀 + 𝑤 = 0

08/07/2017 Dr. Luis Mosquera L. 186


From which 𝜀 + 𝑤 = ∇𝑢

Taken the symmetric part of this equation gives us the strain

1
𝜀 = (∇𝑢 + ∇𝑢𝑇 )
2
1
or 𝜀𝑖𝑗 = (𝑢𝑖,𝑗 + 𝑢𝑗 ,𝑖 )
2

08/07/2017 Dr. Luis Mosquera L. 187


Stretch Ratios

Define the ratio of dx and dX to be the stretch ratio, Λ. For the differential element in the
direction of the unit vector N at P, we write
𝑑𝑥
𝛬 𝑁 =
𝑑𝑋

2
Thus, 𝑑𝑥 = 𝑑𝑥. 𝑑𝑥 = 𝐹. 𝑑𝑋. 𝐹. 𝑑𝑋 = 𝑑𝑋. 𝐹 𝑇 𝐹. 𝑑𝑋
𝑑𝑿 𝑇 𝑑𝑿
𝛬2𝑁 = . 𝐹 𝐹. = 𝑁. 𝐹 𝑇 𝐹. 𝑁
𝑑𝑋 𝑑𝑋
In an analogous way
1 𝑑𝑋
=
𝜆𝑛 𝑑𝑥
1 𝑑𝒙 −1 𝑇 −1 𝑑𝒙 𝑇
= .𝐹 𝐹 . = 𝑛. 𝐹 −1 𝐹 −1 . 𝑛
𝜆2𝑛 𝑑𝑥 𝑑𝑥
08/07/2017 Dr. Luis Mosquera L. 188
For 𝑁 = 𝐼1
𝛬2𝐼1 = 1 + 2𝐸11

And for 𝑛 = 𝑒1
1
= 1 − 2𝑒11
𝜆2𝑒1

The unit extension in any direction N at P. This may be expressed as

𝑑𝑥 − 𝑑𝑋
𝑒 𝑁 = =𝛬 𝑁 −1= 𝑁. 𝐹 𝑇 𝐹. 𝑁 − 1
𝑑𝑋

08/07/2017 Dr. Luis Mosquera L. 189


If
𝑁 = 𝐼1

𝑒 𝐼1 = 𝐼1 . 𝐹 𝑇 𝐹. 𝐼1 − 1 = 1 + 2𝐸11 − 1

For small deformation,


𝐸11 → 𝜀11 = 𝑒 𝐼1

The change in angle between two line elements is given as

𝑁1 . 𝐹 𝑇 𝐹. 𝑁2
𝑐𝑜𝑠𝜃 =
𝛬 𝑁1 𝛬 𝑁2

08/07/2017 Dr. Luis Mosquera L. 190


In a similar fashion,

𝑇
𝑐𝑜𝑠𝛩 = 𝜆 𝑛1 𝜆 𝑛2 𝑛1 . 𝐹 −1 𝐹 −1 . 𝑛2

Example: A homogeneous deformation is given by the mapping equations,

𝑥1 = 𝑋1 − 𝑋2 + 𝑋3

𝑥2 = 𝑋1 + 𝑋2 − 𝑋3

𝑥3 = 𝑋3 − 𝑋1 + 𝑋2

Determine
a) The stretch ratio in the direction of 𝑁1 = 𝐼1 + 𝐼2 / 2

b) The angle θ12 in the deformed configuration between elements that were
originally in the directions of N1 and N2 = I2

08/07/2017 Dr. Luis Mosquera L. 191


Solution
1 −1 1
a) 𝐹 = 1 1 −1
−1 1 1

1
1 1 3 −1 −1 2
𝛬2𝑁1 = 0 −1 3 −1 1 =2
2 2 −1 −1 3 2
0

b) 𝛬2𝑁2 = 3

𝐼1 + 𝐼2 / 2. 𝐹 𝑇 𝐹. 𝐼2 2/ 2
𝑐𝑜𝑠𝜃12 = =
2 3 6

And θ12 = 54.7° . Thus, the original 45° angle is enlarged by 9.7°

08/07/2017 Dr. Luis Mosquera L. 192


The ratio of the deformed volume to the original becomes,

1
𝑑𝑉 𝑑𝑥 𝑑𝑥 2 𝑑𝑥 3
= =Λ 𝑁1 Λ 𝑁2 Λ 𝑁3
𝑑𝑉 0 𝑑𝑋 1 𝑑𝑋 2 𝑑𝑋 3

For the deformation of the example,

𝑁1 = 𝐼1 + 𝐼2 + 𝐼3 / 3 𝛬 𝑁1 = 1

𝑁2 = 𝐼1 − 𝐼2 / 2 𝛬 𝑁2 = 2

𝑁3 = 𝐼1 + 𝐼2 − 2𝐼3 / 6 𝛬 𝑁3 = 2

𝑑𝑉
Using these results, =4
𝑑𝑉 0
08/07/2017 Dr. Luis Mosquera L. 193
Velocity Gradient, Rate of Deformation, Vorticity

Let the velocity field of a continuum given by vi =vi (x,t). The spatial velocity gradient is
defined by
𝜕𝑣𝑖
𝐿𝑖𝑗 =
𝜕𝑥𝑗

𝐿𝑖𝑗 = 𝑑𝑖𝑗 + 𝑤𝑖𝑗

1 𝜕𝑣𝑖 𝜕𝑣𝑗
𝑑𝑖𝑗 = +
2 𝜕𝑥𝑗 𝜕𝑥𝑖

Where, dijo es te arte of deformation tensor

08/07/2017 Dr. Luis Mosquera L. 194


and 1 𝜕𝑣𝑖 𝜕𝑣𝑗
𝑤𝑖𝑗 = −
2 𝜕𝑥𝑗 𝜕𝑥𝑖

Is the vorticity, or spin tensor

08/07/2017 Dr. Luis Mosquera L. 195


Consider two neighboring points p and q. The particle at q has a velocity relative to the
particle at p of
𝜕𝑣𝑖
𝑑𝑣𝑖 = 𝑑𝑥
𝜕𝑥𝑗 𝑗
𝑑𝑣 = 𝐿. 𝑑𝑥

Note that
𝜕𝑣𝑖 𝜕𝑣𝑖 𝜕𝑋𝐴 𝑑 𝜕𝑥𝑖 𝜕𝑋𝐴
= =
𝜕𝑥𝑗 𝜕𝑋𝐴 𝜕𝑥𝑗 𝑑𝑡 𝜕𝑋𝐴 𝜕𝑥𝑗

In symbolic notation

𝐿 = 𝐹 . 𝐹 −1

𝐹 = 𝐿. 𝐹

08/07/2017 Dr. Luis Mosquera L. 196


Consider next the stretch ratio 𝑑𝑥
𝛬=
𝑑𝑋

𝑛𝛬 = 𝐹. 𝑁

If we take the material derivative of this equation,

𝑛𝛬 + 𝑛𝛬 = 𝐹 . 𝑁 = 𝐿. 𝐹. 𝑁 = 𝐿. 𝑛𝛬
𝑛. 𝑛 = 1 𝑎𝑛𝑑 𝑠𝑜 𝑛. 𝑛 = 0

resulting in 𝛬
= 𝑛. 𝐿. 𝑛 = 𝑑𝑖𝑗 𝑛𝑖 𝑛𝑗
𝛬

Which represents the rate of stretching per unit stretch of the element that originated
in the direction of N, and is in the direction of n of the current configuration.

08/07/2017 Dr. Luis Mosquera L. 197


Recall that 2𝐸 = 𝐹 𝑇 𝐹 − 𝐼

Therefore,
2𝐸 = 𝐹 𝑇 . 𝐿𝑇 + 𝐿 . 𝐹 = 𝐹 𝑇 . 2𝐷. 𝐹

(𝑝)
since 𝑛𝑖 = 𝑤𝑖𝑗 𝑛𝑗 and because a unit vector can change only in direction, then
wij gives the rate of change in direction of the principal axes
of D. Hence the names, vorticity or spin given to W.

1
Additionally, 𝑤 = 𝑐𝑢𝑟𝑙 𝑣 , is called vorticity vector
2

08/07/2017 Dr. Luis Mosquera L. 198


Material Derivative of Line Elements, Areas, Volumes

𝑑𝑥 = 𝐹. 𝑑𝑋

𝑑𝑥 = 𝐹 . 𝑑𝑋 = 𝐿. 𝐹. 𝑑𝑋 = 𝐿. 𝑑𝑥

𝑑𝑥𝑖 = 𝑣𝑖,𝑗 𝑑𝑥𝑗

𝑑𝑥. 𝑑𝑥 = 2𝑑𝑥. 𝑑𝑥 = 𝑑𝑥. 2𝐿. 𝑑𝑥 = 𝑑𝑥. 2𝐷. 𝑑𝑥

08/07/2017 Dr. Luis Mosquera L. 199


Consider the plane area defined in the reference configuration by the differential
line elements
(1) (2)
𝑑𝑋𝐴 𝑎𝑛𝑑 𝑑𝑋𝐴

The parallelogram area dS° may be represented by

(1) (2)
𝑑𝑆𝐴0 = ϵ𝐴𝐵𝐶 𝑑𝑋𝐵 𝑑𝑋𝐶

As a result of the motion, this area is carried into the current area dSi
(1) (2)
𝑑𝑆𝑖 = ϵ𝑖𝑗𝑘 𝑑𝑥𝑗 𝑑𝑥𝑘

𝑑𝑆𝑞 = 𝑋𝐴,𝑞 𝐹 𝑑𝑆𝐴0

𝑑𝑆. 𝐹 = 𝐹 𝑑𝑆 0

Also,
𝑑𝑉 = 𝐹 𝑑𝑉 0

08/07/2017 Dr. Luis Mosquera L. 200


P18.- Let a displacement field be given by

1
𝑢1 = 𝑋3 − 𝑋2
4
1
𝑢2 = 𝑋1 − 𝑋3
4
1
𝑢3 = 𝑋2 − 𝑋1
4

Determine the volume ratio dV/dV°

08/07/2017 Dr. Luis Mosquera L. 201


P19.- Given the deformation expressed by

𝑥1 = 𝑋1 + 𝐴𝑋22
𝑥2 = 𝑋2
𝑥3 = 𝑋3 − 𝐴𝑋22

Where A is a constant (not necessarily small), determine the finite strain tensors E
and ε, and show that if the displacements are small so that x ≈ X, and if squares of A
may be neglected, both tensors reduce to the infinitesimal strain tensor.

08/07/2017 Dr. Luis Mosquera L. 202


P20.- Let the deformation

𝑥1 = 𝑎1 (𝑋1 + 2𝑋2 )
𝑥2 = 𝑎2 𝑋2
𝑥3 = 𝑎3 𝑋3

Where a1, a2 and a3 are constants be applied to the unit cube of material shown in
the sketch. Determine
a) The deformed length l of diagonal OC
b) The angle between edges OA and OG after deformation,
c) The conditions which the constants must satisfy for the deformation to be
possible if
i) The material is incompressible,
ii) The angle between elements OC and OB is to remain unchanged

08/07/2017 Dr. Luis Mosquera L. 203


Material derivatives of line, surface and volume integrals

Let a property of the collection of particles occupying the current volume V be


represented by the integral

𝑃𝑖𝑗 … 𝑡 = 𝑃𝑖𝑗∗ … 𝑥, 𝑡 𝑑𝑉
𝑉

The material derivative of this property is given in both spatial and material form by

𝑑 𝑑
𝑃𝑖𝑗 … 𝑡 = 𝑃𝑖𝑗∗ … 𝑥, 𝑡 𝑑𝑉 = 𝑃𝑖𝑗∗ … 𝑥 𝑋, 𝑡 , 𝑡 𝐽𝑑𝑉 0
𝑑𝑡 𝑉 𝑑𝑡 𝑉0

This equation maybe written,

[𝑃𝑖𝑗∗… 𝑋, 𝑡 𝐽] 𝑑𝑉 0 = (𝑃𝑖𝑗∗ … + 𝑣𝑘,𝑘 𝑃𝑖𝑗∗ … ) 𝐽𝑑𝑉 0


𝑉0 𝑉0

Ref. P. 148
08/07/2017 Dr. Luis Mosquera L. 204
Which upon application of the divergence theorem becomes

𝜕𝑃𝑖𝑗∗ …
𝑃𝑖𝑗 … 𝑡 = 𝑑𝑉 + 𝑣𝑘 𝑃𝑖𝑗∗ … 𝑛𝑘 𝑑𝑆
𝑉 𝜕𝑡 𝑆

This equation gives the time rate of change of the property Pij… as the sum of the
amount created in the volume V, plus the amount entering through the bounding
surface S, and is often spoken of as the transport theorem.

Conservation of Mass, Continuity Equation

The mass m of the entire body is given by

𝑚= 𝜌 𝑥, 𝑡 𝑑𝑉
𝑉

08/07/2017 Dr. Luis Mosquera L. 205


The mass of the body in the referencial configuration

𝑚= 𝜌0 𝑋, 𝑡 𝑑𝑉 0
𝑉0

The law of conservation of mass asserts that the mass of a body, or of any portion of
the body, is invariant under motion. Thus,

𝑑
𝑚= 𝜌 𝑥, 𝑡 𝑑𝑉 = (𝜌 + 𝜌𝑣𝑖,𝑖 ) 𝑑𝑉 = 0
𝑑𝑡 𝑉 𝑉

Therefore, 𝜌 + 𝜌𝑣𝑖,𝑖 = 0

Which is known as the continuity equation in Eulerian form. But the material
derivative of ρ is,
𝜕𝜌 𝜕𝜌
𝜌= + 𝑣𝑖
𝜕𝑡 𝜕𝑥𝑖

08/07/2017 Dr. Luis Mosquera L. 206


So that, 𝜕𝜌
+ 𝜌𝑣𝑖 , 𝑖 = 0
𝜕𝑡
If 𝜌 = 0, the material is said to be incompressible, and thus

𝑣𝑖,𝑖 = 0 𝑜𝑟 𝑑𝑖𝑣 𝒗 = 0

Since the law of conservation of mass requires the mass to be the same in all
configurations, thus we have

𝜌 𝑥 𝑋, 𝑡 , 𝑡 𝑑𝑉 = 𝜌 𝑋, 𝑡 𝐽𝑑𝑉 0
𝑉 𝑉0

and so in the material description,

𝜌𝐽 = 𝜌0

𝜌𝐽 = 0
This equation is said the Lagrangian, or material, form of the continuity equation
08/07/2017 Dr. Luis Mosquera L. 207
Linear Momentum Principle, Equations of Motion

Let a material continuum body having a current volume V and bounding surface S be
subjected to surface traction 𝑡𝑖 𝑛 and distributed body forces ρbi . In addition, let the
body be in motion under the velocity field vi = vi(x,t).

The linear momentum of the body is

𝑃𝑖 𝑡 = 𝜌𝑣𝑖 𝑑𝑉
𝑉

and the principle of linear momentum


states that
𝑑
𝜌𝑣𝑖 𝑑𝑉 = 𝑡𝑖 𝑛 𝑑𝑆 + 𝜌𝑏𝑖 𝑑𝑉
𝑑𝑡 𝑉 𝑆 𝑉

08/07/2017 Dr. Luis Mosquera L. 208


Where,
𝜌𝑣𝑖 − 𝑡𝑗𝑖 ,𝑗 − 𝜌𝑏𝑖 𝑑𝑉 = 0
𝑉

again, V is arbitrary and so the integrand must vanish, and we obtain,

𝜌𝑣𝑖 − 𝑡𝑗𝑖 ,𝑗 − 𝜌𝑏𝑖 = 0

which are known as the local equations of motion in Eulerian form.

When the velocity field is zero, or constant, the equations of motion reduce to the
equilibrium equations,

𝑡𝑗𝑖 ,𝑗 + 𝜌𝑏𝑖 = 0

08/07/2017 Dr. Luis Mosquera L. 209


Moment of Momentum (Angular Momentum)
Principle

The principle of angular momentum states that the time rate of change of the
moment of momentum of a body with respect to a given point is equal to the
moment of the surface and body forces with respect to that point.

if we take the origin as the point of


reference, the angular momentum
principle has the mathematical form

𝑑 (𝑛 )
𝜀𝑖𝑗𝑘 𝑥𝑗 𝜌𝑣𝑘 𝑑𝑉 = 𝜀𝑖𝑗𝑘 𝑥𝑗 𝑡𝑘 𝑑𝑆 + 𝜀𝑖𝑗𝑘 𝑥𝑗 𝜌𝑏𝑘 𝑑𝑉
𝑑𝑡 𝑉 𝑆 𝑉

08/07/2017 Dr. Luis Mosquera L. 210


results in

𝜀𝑖𝑗𝑘 𝑥𝑗 𝜌𝑣𝑘 − 𝑡𝑞𝑘 ,𝑞 − 𝜌𝑏𝑘 − 𝑡𝑗𝑘 𝑑𝑉 = 0


𝑉

which reduces to

𝜀𝑖𝑗𝑘 𝑡𝑘𝑗 𝑑𝑉 = 0
𝑉

Again, with V arbitrary, the integrand must vanish so that

𝜀𝑖𝑗𝑘 𝑡𝑘𝑗 = 0

which by direct expansion demonstrates that 𝑡𝑘𝑗 = 𝑡𝑗𝑘 , and the stress tensor
is symmetric.

08/07/2017 Dr. Luis Mosquera L. 211


Law of Conservation of Energy,

the material time derivative of the kinetic plus internal energies is equal to the
sum of the rate of work of the surface and body forces, plus all other energies that
enter or leave the body per unit time. Other energies may include, for example,
thermal, electrical, magnetic, or chemical energies.
By definition, the kinetic energy of the material occupying an arbitrary volume V
of the body is

1 1
𝐾 𝑡 = 𝜌𝒗. 𝒗𝑑𝑉 = 𝜌𝑣𝑖 𝑣𝑖 𝑑𝑉
2 𝑉 2 𝑉

Also, the mechanical power, or rate of work of the body and surface forces shown
in the figure is defined by the scalar

𝑃 𝑡 = 𝑡𝑖 𝑛 𝑣𝑖 𝑑𝑆 + 𝜌𝑏𝑖 𝑣𝑖 𝑑𝑉
𝑆 𝑉

08/07/2017 Dr. Luis Mosquera L. 212


Consider now the material derivative of the kinetic energy integral

𝐾= 𝜌 𝑣𝑖 𝑣𝑖 𝑑𝑉 = 𝑣𝑖 𝑡𝑗𝑖 ,𝑗 + 𝜌𝑏𝑖 𝑑𝑉
𝑉 𝑉

which, if we convert the middle term by the divergence theorem and make
use of the decomposition

𝑣𝑖,𝑗 = 𝑑𝑖𝑗 + 𝑤𝑖𝑗 , may be written

𝐾= 𝜌𝑏𝑖 𝑣𝑖 𝑑𝑉 + 𝑡𝑖 𝑛 𝑣𝑖 𝑑𝑆 − 𝑡𝑖𝑗 𝑑𝑖𝑗 𝑑𝑉


𝑉 𝑆 𝑉

Therefore,

𝐾+𝑆 =𝑃

08/07/2017 Dr. Luis Mosquera L. 213


where the integral

𝑆= 𝑡𝑖𝑗 𝑑𝑖𝑗 𝑑𝑉 = 𝑇: 𝐷 𝑑𝑉
𝑉 𝑉

is known as the stress work

The balance of mechanical energy shows that, of the total work done by the
external forces, a portion goes toward increasing the kinetic energy, and the
remainder appears as work done by the internal stresses.

In the special situation when

𝑑
𝑆=𝑈= 𝜌𝑢𝑑𝑉 = 𝜌𝑢 𝑑𝑉
𝑑𝑡 𝑉 𝑉

where U is called the internal energy and u the specific internal energy, or energy
density (per unit mass),then

08/07/2017 Dr. Luis Mosquera L. 214


𝑑 1 𝑛
𝜌 𝑣𝑖 𝑣𝑖 + 𝑢 𝑑𝑉 = 𝜌𝑏𝑖 𝑣𝑖 𝑑𝑉 = 𝑡𝑖 𝑣𝑖 𝑑𝑆
𝑑𝑡 𝑉 2 𝑉 𝑆

or, briefly,
𝐾+𝑈=𝑃

For a thermomechanical continuum, we represent the rate at which


thermal energy is added to a body by

𝑄= 𝜌𝑟𝑑𝑉 − 𝑞𝑖 𝑛𝑖 𝑑𝑆
𝑉 𝑆

The scalar field r specifies the rate at which heat per unit mass is produced by
internal sources and is known as the heat supply. The vector qi, called the
heat flux vector, is a measure of the rate at which heat is conducted into the
body per unit area per unit time across the element of surface dS whose
outward normal is ni .
08/07/2017 Dr. Luis Mosquera L. 215
The heat flux qi is often assumed to obey Fourier’s law of heat conduction;

𝑞𝑖 = −𝑘𝜃,𝑖 𝑜𝑟 𝒒 = −𝑘∇𝜃

where κ is the thermal conductivity and 𝜃,𝑖 is the temperature gradient.

With the addition of the thermal energy consideration, the complete energy
balance takes the form
𝐾+𝑈 =𝑃+𝑄
or, when written out in detail,

𝜌𝑢 − 𝑡𝑖𝑗 𝑑𝑖𝑗 − 𝜌𝑟 + 𝑞𝑖,𝑖 = 0

which is called the energy equation.

08/07/2017 Dr. Luis Mosquera L. 216


Entropy and the Clausius-Duhem Equation

The conservation of energy as formulated in Section 5.7 is a statement of the


interconvertibility of heat and work. However, there is not total interconvertibility
for irreversible processes. For instance, the case of mechanical work being
converted to heat via friction is understood, but the converse does not hold. That
is, heat cannot be utilized to directly generate work.
This, of course, is the motivation for the second law of thermodynamics.

In continuum mechanics, a statement of the second law is made to place


restrictions on continua. However, in the case of continuum mechanics the
restrictions are placed on the material response functions called constitutive
responses.

08/07/2017 Dr. Luis Mosquera L. 217


At any given state for the continuum there are various quantities that affect the
internal energy. These might be the volume of an ideal gas or the components of
the deformation gradient of a solid. In the case of the deformation gradient, the
nine components represent a deformation in the body that is storing energy. The
collection of these parameters is called the thermodynamic substate and will be
denoted by 𝑣1 𝑣2 … . . 𝑣𝑛

While the thermodynamic substate influences the internal energy of the body
it does not completely define it. Assume that the substate plus an additional
independent scalar parameter, η, is sufficient to define the internal energy. This
definition may be made in the form of

𝑢 = 𝑓 𝜂, 𝑣1 , 𝑣2 , … 𝑣𝑛

which is often referred to as the caloric equation of state. Parameter η is called


the specific entropy.

08/07/2017 Dr. Luis Mosquera L. 218


Temperature is the result of the change in internal energy with respect to
entropy
𝜕𝑢
𝜃=
𝜕𝜂

partial differentiation of the internal energy with respect to the thermodynamic


substate variables results in thermodynamic tensions
𝜕𝑢
𝜏𝑎 =
𝜕𝑣𝑎
We write a differential form of the internal energy as follows:

𝑑𝑢 = 𝜃𝑑𝜂 + 𝜏𝑎 𝑑𝑣𝑎
𝑎

Assuming that all the functions defined in the section are continuously differentiable
as many times as necessary, it is possible to solve for entropy in terms of temperature

08/07/2017 Dr. Luis Mosquera L. 219


𝜂 = 𝜂 𝜃, 𝑣𝑎

This result may be substituted into the caloric equation of state to yield internal
energy as a function of temperature and substate parameters

𝑢 = 𝑢 𝜃, 𝑣𝑎
Using this result allows the definition of the thermal equations of state

𝜏𝑎 = 𝜏𝑎 𝜃, 𝑣𝑎
The principles of thermodynamics are often posed in terms of thermodynamic
potentials which may be defined as follows:

08/07/2017 Dr. Luis Mosquera L. 220


internal energy u

Free energy 𝜓 = 𝑢 − 𝜂𝜃

Enthalpy 𝜒=𝑢− 𝑎 𝜏𝑎 𝑣𝑎

Free enthalpy 𝜁 = 𝜒 − 𝜂𝜃

These potentials are related through the relationship


𝑢−𝜓+𝜁−𝜒 =0

When considered for a portion P of the body, the total entropy is given as

𝐻= 𝜌𝜂𝑑𝑉
𝑃

and the entropy production in the portion P is given by

08/07/2017 Dr. Luis Mosquera L. 221


Γ= 𝜌𝛾𝑑𝑉
𝑃

where the scalar γ is the specific entropy production. The second law can be stated as
follows: the time rate-of-change in the entropy equals the change in entropy due to
heat supply, heat flux entering the portion, plus the internal entropy production. For
a portion P of the body, this is written as

𝑑 𝜌𝑟 𝑞𝑖 𝑛𝑖
𝜌𝜂𝑑𝑉 = 𝑑𝑉 − 𝑑𝑆 + 𝜌𝛾𝑑𝑉
𝑑𝑡 𝑃 𝑃 𝜃 𝜕𝑃 𝜃 𝑃

The entropy production is always positive, which leads to a statement of the


second law in the form of the Clausius-Duhem inequality

1
𝜌𝜃𝜂 − 𝜌𝑟 + 𝑞𝑖,𝑖 − 𝑞𝑖 𝜃,𝑖 ≥ 0
𝜃

08/07/2017 Dr. Luis Mosquera L. 222


Restrictions on Elastic Materials by the
Second Law of Thermodynamics

For a continuum body B having material points X a thermodynamic process is


described by eight functions of the material point and time. These functions
would be as follows:

1. Spatial position 𝑥𝑖 = 𝑋𝑖 𝑋, 𝑡
2. Stress tensor 𝜎𝑖𝑗 = 𝜎𝑖𝑗 𝑋, 𝑡
3. Body force per unit mass bi = bi(X,t)
4. Specific internal energy u = u(X,t)
5. Heat flux vector qi = qi(X,t)
6. Heat supply per unit mass r = r(X,t)
7. Specific entropy η = η(X,t)
8. Temperature (always positive) θ = θ(X,t)

08/07/2017 Dr. Luis Mosquera L. 223


A set of these eight functions which are compatible with the balance of linear
momentum and the conservation of energy makes up a thermodynamic process.

Invariance

Invariance plays another important role in continuum mechanics. Requiring a


continuum to be invariant with regards to reference frame. There are two basic
methods for examining invariance of constitutive response functions: material
frame indifference and superposed rigid body motion.

In the first, a continuum body’s response to applied forces or prescribed motion must
be the same as observed from two different reference frames. The body and the
applied forces remain the same; only the observer’s reference frame changes.

08/07/2017 Dr. Luis Mosquera L. 224


the superposed motion can be written as

𝑝𝑖+ = 𝑐𝑖 𝑡 + 𝑄𝑖𝑚 𝑡 𝑝𝑚

+
where vectors 𝑝𝑖 𝑎𝑛𝑑 𝑝𝑚

are defined as shown in figure.


Here, 𝑄𝑖𝑚 is simply the matrix of
the direction cosines between
Ox1x2x3 and 𝑂+𝑥1+𝑥2+𝑥3+

08/07/2017 Dr. Luis Mosquera L. 225


Restrictions on Constitutive
Equations from Invariance

Under a superposed rigid body motion scalars are unaffected, allowing the
following to be written

𝑢+ = 𝑢, 𝜂+ = 𝜂, 𝜃+ = 𝜃

ψ, χ, and ζ would be unaffected by the superposed rigid body motion. The remaining
quantities of the response functions and their independent variables are affected in
different ways from the superposed rigid body motion.
Under a superposed rigid body motion, these functions transform as follows :

+
𝐹𝑖𝐴 = 𝑄𝑖𝑗 𝐹𝑗𝐴
+
𝐹𝑖𝐴 = 𝑄𝑖𝑗 𝐹𝑗𝐴 + Ω𝑖𝑗 𝑄𝑗𝑘 𝐹𝑘𝐴

08/07/2017 Dr. Luis Mosquera L. 226


𝐿+𝑖𝑚 = 𝑄𝑖𝑗 𝐿𝑗𝑛 𝑄𝑛𝑚 + Ω𝑖𝑚

𝑡𝑖𝑗+ = 𝑄𝑖𝑘 𝑡𝑘𝑙 𝑄𝑙𝑗

𝑞𝑖+ = 𝑄𝑖𝑗 𝑞𝑗

𝑔𝑖+ = 𝑄𝑖𝑗 𝑔𝑗

Constitutive equations are objective if and only if they transform under a


superposed rigid body motion as follows

+
𝑢+ 𝜂+, 𝐹𝑖𝐴 = 𝑢(𝜂, 𝐹𝑖𝐴 )
+
𝜃 + 𝜂+, 𝐹𝑖𝐴 = 𝑢(𝜂, 𝐹𝑖𝐴 )
𝑡𝑖𝑗+ 𝜂+, 𝐹𝑚𝐴
+
= 𝑄𝑖𝑘 𝑡𝑘𝑙 (𝜂, 𝐹𝑚𝐴 )𝑄𝑙𝑘

𝑞𝑖+ 𝜂+, 𝐹𝑖𝐴


+
, 𝑔𝑖+, 𝐿+𝑘𝑙 = 𝑄𝑖𝑚 𝑞𝑚 (𝜂, 𝐹𝑖𝐴 , 𝑔𝑖 , 𝐿𝑘𝑙 )

08/07/2017 Dr. Luis Mosquera L. 227


Constitutive Equations

The global balance laws and resulting field equations developed


earlier in this chapter are applicable to all continuous media, but say
nothing about the response of specific materials to force or
temperature loadings. To fill this need, we introduce the so-called
constitutive equations, which specify the mechanical and thermal
properties of particular materials based upon their internal
constitution.

08/07/2017 Dr. Luis Mosquera L. 228


A brief listing of some well-known constitutive equations is as follows:

(a) the stress-strain equations for a linear elastic solid assuming infinitesimal
strains,
𝑡𝑖𝑗 = 𝐶𝑖𝑗𝑘𝑚 𝜀𝑘𝑚

Where the Cijkm are the elastic constants representing the properties of the
body.

For isotropic behavior, 𝑡𝑖𝑗 = 𝜆𝛿𝑖𝑗 𝜀𝑘𝑘 + 2𝜇𝜀𝑖𝑗

in which λ and μ are coefficients that express the elastic properties of the
material.

08/07/2017 Dr. Luis Mosquera L. 229


(b) the linear viscous fluid,

𝜏𝑖𝑗 = 𝐾𝑖𝑗𝑚𝑛 𝑑𝑚𝑛

where τij is the shearing stress in the fluid and the constants Kijmn represent its
viscous properties. For a Newtonian fluid,

𝜏𝑖𝑗 = 𝜆∗ 𝛿𝑖𝑗 𝑑𝑘𝑘 + 2𝜇∗ 𝑑𝑖𝑗


where λ* and μ* are viscosity coefficients.

(c) plastic stress-strain equation,

𝑑𝜖𝑖𝑗𝑃 = 𝑆𝑖𝑗 𝑑𝜆

𝑃
Where 𝑑𝜖𝑖𝑗 is the plastic strain increment, Sij the deviator stress, and dλ a
proportionality constant.

08/07/2017 Dr. Luis Mosquera L. 230


In formulating a well-posed problem in continuum mechanics, we need the
field equations together with whatever equations of state are necessary, plus
the appropriate constitutive equations and boundary conditions. As a point of
reference we list again, as a group, the important field equations in indicial
notation:

(a) the continuity equation


𝜕𝜌
+ 𝜌𝑣𝑖 , 𝑖 = 0
𝜕𝑡

(b) the equations of motion

𝜌𝑣𝑖 − 𝑡𝑗𝑖 ,𝑗 − 𝜌𝑏𝑖 = 0

(c) the energy equation

𝜌𝑢 − 𝑡𝑖𝑗 𝑑𝑖𝑗 − 𝜌𝑟 + 𝑞𝑖,𝑖 = 0

08/07/2017 Dr. Luis Mosquera L. 231


If we assume the body forces bi and distributed heat sources r are
prescribed, the above collection consists of five independent equations
involving fourteen unknowns, namely, ρ, vi , σij , qi , and u. In addition, in a
non-isothermal situation, the entropy η and temperature field θ = θ(x,t)
have to be taken into consideration. For the isothermal theory, eleven
equations are needed in conjunction with the five field equations listed
above. Of these, six are constitutive equations, three are temperature-heat
conduction equations (Fourier’s law), and two are equations of state.

08/07/2017 Dr. Luis Mosquera L. 232


P21.- Given the velocity field
𝑣1 = 𝑎𝑥1 − 𝑏𝑥2

𝑣2 = 𝑏𝑥1 + 𝑎𝑥2

𝑣3 = 𝑐 𝑥12 + 𝑥22

where a, b, and c are constants, determine


(a) whether or not the continuity equation is satisfied
(b) whether the motion is isochoric.

08/07/2017 Dr. Luis Mosquera L. 233


P22.- For a certain contiuum at rest, the stress is given by

𝑡𝑖𝑗 = −𝑝0 𝛿𝑖𝑗

where p0 is a constant. Use the continuity equation to show that for this case
the stress power may be expressed as

𝑝0 𝜌
𝑡𝑖𝑗 𝑑𝑖𝑗 =
𝜌

08/07/2017 Dr. Luis Mosquera L. 234


P23.-Show that, for a rigid body rotation about the origin, the kinetic energy
integral

1 1
𝐾 𝑡 = 𝜌𝒗. 𝒗𝑑𝑉 = 𝜌𝑣𝑖 𝑣𝑖 𝑑𝑉
2 𝑉 2 𝑉

reduces to the form given in rigid body dynamics, that is,

1
𝐾 = 𝑤𝑖 𝑤𝑖 𝐼𝑖𝑗
2

where Iij is the inertia tensor

08/07/2017 Dr. Luis Mosquera L. 235


P24.- Starting with Equation

𝑡𝑖𝑗 = 𝜆𝛿𝑖𝑗 𝜀𝑘𝑘 + 2𝜇𝜀𝑖𝑗

for isotropic elastic behavior, show that

𝑡𝑖𝑖 = (3𝜆 + 2𝜇)𝜀𝑖𝑖

and, using this result, deduce that

1 𝜆
𝜀𝑖𝑗 = 𝑡 − 𝛿 𝑡
2𝜇 𝑖𝑗 3𝜆 + 2𝜇 𝑖𝑗 𝑘𝑘

08/07/2017 Dr. Luis Mosquera L. 236


Linear Elasticity

Elastic behavior is characterized by the following two conditions:


(1) Where the stress in a material is a unique function of the strain, and
(2) where the material has the property for complete recovery to a
“natural” shape upon removal of the applied forces.

Symbolically, we write the constitutive equation for elastic behavior in its


most general form as
𝑇 = 𝐺(𝜀)

Where, G is a symmetric tensor-valued function. .Also, we acknowledge that


elastic behavior may be linear or non-linear. However, for the response function
G we consider only that case for which the stress is a linear function of strain.

08/07/2017 Dr. Luis Mosquera L. 237


Uniaxial loading-unloading stress-strain curves for (a) linear elastic; (b)
nonlinear elastic; and (c) inelastic behavior.

Also, we assume that, in the deformed material, the displacement gradients


are everywhere small compared with unity. Thus, the distinction between the
Lagrangian and Eulerian descriptions is negligible.

08/07/2017 Dr. Luis Mosquera L. 238


1
𝜀𝑖𝑗 = (𝑢𝑖,𝑗 + 𝑢𝑗 ,𝑖 )
2

Within the context of the above assumptions, we write the constitutive


equation for linear elastic behavior as

𝑡𝑖𝑗 = 𝐶𝑖𝑗𝑘𝑚 𝜀𝑘𝑚

or 𝑇 = 𝐶𝜀
where the tensor of elastic coefficients Cijkm has 34 = 81 components. However,
due to the symmetry of both the stress and strain tensors, it is clear that

𝐶𝑖𝑗 𝑘𝑚 = 𝐶𝑗 𝑖𝑘𝑚 = 𝐶𝑖𝑗𝑚𝑘

which reduces the 81 possibilities to 36 distinct coefficients at most.

08/07/2017 Dr. Luis Mosquera L. 239


The tensor character of C : expressed in a rotated (primed) coordinate system, it has
the form

𝑡′𝑖𝑗 = 𝐶′𝑖𝑗𝑘𝑚 𝜀′𝑘𝑚

By the transformation laws for second-order tensors,

𝑡′𝑖𝑗 = 𝑎𝑖𝑞 𝑎𝑗𝑠 𝑡𝑞𝑠 = 𝑎𝑖𝑞 𝑎𝑗𝑠 𝐶𝑞𝑠𝑘𝑚 𝜀𝑘𝑚

= 𝑎𝑖𝑞 𝑎𝑗𝑠 𝐶𝑞𝑠𝑘𝑚 𝑎𝑝𝑘 𝑎𝑛𝑚 𝜀′𝑝𝑛

which provides the result

𝐶′𝑖𝑗 𝑘𝑚 = 𝑎𝑖𝑞 𝑎𝑗𝑠 𝑎𝑝𝑘 𝑎𝑛𝑚 𝐶𝑞𝑠𝑘𝑚

that is, the transformation rule for a fourth-order Cartesian tensor.

08/07/2017 Dr. Luis Mosquera L. 240


In general, the Cijkm coefficients may depend upon temperature, but here we
assume adiabatic (no heat gain or loss) and isothermal (constant temperature)
conditions.
We also shall ignore strain-rate effects and consider the components Cijkm to be
at most a function of position.
If the elastic coefficients are constants, the material is said to be homogeneous.

It is convenient to define

𝑡11 = 𝑡1 , 𝑡23 = 𝑡32 = 𝑡4


𝑡22 = 𝑡2 , 𝑡13 = 𝑡31 = 𝑡5
𝑡33 = 𝑡3 , 𝑡12 = 𝑡21 = 𝑡6

08/07/2017 Dr. Luis Mosquera L. 241


and
𝜀11 = 𝜖1 , 2𝜀23 = 2𝜀32 = 𝜖4
𝜀22 = 𝜖2 , 2𝜀13 = 2𝜀31 = 𝜖5
𝜀33 = 𝜖3 , 2𝜀12 = 2𝜀21 = 𝜖6

From these definitions, Hooke’s law is now written

𝑡𝛼 = 𝒞𝛼𝛽 𝜖𝛽

with Greek subscripts having a range of six. In matrix form this equation appears
as
𝑡1 𝒞11 𝒞12 𝒞13 𝒞14 𝒞15 𝒞16 𝜖1
𝑡2 𝒞21 𝒞22 𝒞23 𝒞24 𝒞25 𝒞26 𝜖2
𝑡3 𝒞31 𝒞32 𝒞33 𝒞34 𝒞35 𝒞36 𝜖3
𝑡4 = 𝒞41 𝒞42 𝒞43 𝒞44 𝒞45 𝒞46 𝜖4
𝑡5 𝒞51 𝒞52 𝒞53 𝒞54 𝒞55 𝒞56 𝜖5
𝑡6 𝒞61 𝒞62 𝒞63 𝒞64 𝒞65 𝒞66 𝜖6

08/07/2017 Dr. Luis Mosquera L. 242


The array of the 36 constants Cαβ does not constitute a tensor.

If thermal effects are neglected, the energy balance equation

𝜌𝑢 − 𝑡𝑖𝑗 𝑑𝑖𝑗 − 𝜌𝑟 + 𝑞𝑖,𝑖 = 0

is reduced to the form


1 1 𝜕𝑣𝑖 𝜕𝑣𝑗
𝑢 = 𝑡𝑖𝑗 𝑑𝑖𝑗 ; 𝑑𝑖𝑗 = +
𝜌 2 𝜕𝑥𝑗 𝜕𝑥𝑖

which for small-deformation theory becomes

1
𝑢 = 𝑡𝑖𝑗 𝜀𝑖𝑗
𝜌

The internal energy u in these equations is purely mechanical and is called the
strain energy (per unit mass).

08/07/2017 Dr. Luis Mosquera L. 243


Recall now that, by the continuity equation in Lagrangian form, ρo = ρJ and also
that to the first order of approximation
𝜕𝑢𝑖 𝜕𝑢𝑖
𝐽 = det 𝐹 = 𝑑𝑒𝑡 𝛿𝑖𝐴 + =1+
𝜕𝑋𝐴 𝜕𝑋𝐴

Therefore, from our assumption of small displacement gradients, namely


𝝏𝒖𝒊 /𝝏𝑿𝑨 ≪ 𝟏, we may take J ≈1 in the continuity equation to give ρ = ρ0, a constant.

For elastic behavior under these assumptions, the strain energy is a function of
the strain components only, and we write

𝜕𝑢
𝑢= 𝜀
𝜕𝜀𝑖𝑗 𝑖𝑗
1
also 𝑢 = 𝑡𝑖𝑗 𝑑𝑖𝑗
𝜌

1 𝜕𝑢
Therefore, 𝑡 =
𝜌 𝑖𝑗 𝜕𝜀𝑖𝑗
08/07/2017 Dr. Luis Mosquera L. 244
The strain energy density, W (strain energy per unit volume) is defined by

𝑊 = 𝜌0 𝑢
From it follows that
𝜕𝑢 𝜕𝑊
𝑡𝑖𝑗 = 𝜌 =
𝜕𝜀𝑖𝑗 𝜕𝜀𝑖𝑗

A material defined in this way is called a hyperelastic material.

Expanding W about the origin, we have

𝜕𝑊 𝜕𝑊(0) 𝜕 2 𝑊(0)
𝑡𝑖𝑗 = = + 𝜀 +⋯
𝜕𝜀𝑖𝑗 𝜕𝜀𝑖𝑗 𝜕𝜀𝑖𝑗 𝜕𝜀𝑘𝑚 𝑘𝑚

It is customary to assume that 𝑡𝑖𝑗 = 0 𝑤ℎ𝑒𝑛 𝜀𝑖𝑗 = 0, thus, we may express the
linear elastic constitutive equation as

08/07/2017 Dr. Luis Mosquera L. 245


𝜕 2 𝑊(0)
𝑡𝑖𝑗 = 𝜀 = 𝐶𝑖𝑗𝑘𝑚 𝜀𝑘𝑚
𝜕𝜀𝑖𝑗 𝜕𝜀𝑘𝑚 𝑘𝑚

So that
𝐶𝑖𝑗𝑘𝑚 = 𝐶𝑘𝑚 𝑖𝑗

Thus, the existence of a strain energy function reduces the number of distinct
components of Cijkm from 36 to 21.

Assuming a linear stress-strain relation, we may now write

1 1
𝑊 𝜀𝑖𝑗 = 𝐶𝑖𝑗𝑘𝑚 𝜀𝑖𝑗 𝜀𝑘𝑚 = 𝑡𝑖𝑗 𝜀𝑖𝑗
2 2
Also
1 1
𝑊 𝜖𝛼 = 𝒞𝛼𝛽 𝜖𝛼 𝜖𝛽 = 𝑡𝛼 𝜖𝛼
2 2

08/07/2017 Dr. Luis Mosquera L. 246


Hooke’s Law for Isotropic Media,
Elastic Constants

If the coeficients Cijkm are the same in every set of reference axes at any point
for a given situation, we call it an isotropic elastic material.

Isotropy requires the elastic tensor C to be a fourth-order isotropic tensor. In


general, an isotropic tensor is defined as one whose components are unchanged by
any orthogonal transformation from one set of Cartesian axes to another.

The unit tensor I, having Kronecker deltas as components, and any scalar multiple of
I are the only second-order isotropic tensors.

The only nontrivial third-order isotropic tensor is the permutation symbol. The
most general fourth-order isotropic tensor may be shown to have a form in terms
of Kronecker deltas which we now introduce as the prototype for C, namely,

08/07/2017 Dr. Luis Mosquera L. 247


𝐶𝑖𝑗𝑘𝑚 = 𝜆𝛿𝑖𝑗 𝛿𝑘𝑚 + 𝜇 𝛿𝑖𝑘 𝛿𝑗𝑚 + 𝛿𝑖𝑚 𝛿𝑗𝑘
+ 𝛽 𝛿𝑖𝑘 𝛿𝑗𝑚 − 𝛿𝑖𝑚 𝛿𝑗𝑘

where λ, μ, and β are scalars.

As 𝐶𝑖𝑗𝑘𝑚 = 𝐶𝑗𝑖𝑘𝑚 = 𝐶𝑖𝑗𝑚𝑘

interchanging i and j in the expression

𝛽 𝛿𝑖𝑘 𝛿𝑗𝑚 − 𝛿𝑖𝑚 𝛿𝑗𝑘 = 𝛽 𝛿𝑗𝑘 𝛿𝑖𝑚 − 𝛿𝑗 𝑚 𝛿𝑖𝑘

see that β = – β and, consequently, β = 0.

Thus,

𝑡𝑖𝑗 = 𝜆𝛿𝑖𝑗 𝛿𝑘𝑚 + 𝜇 𝛿𝑖𝑘 𝛿𝑗𝑚 + 𝛿𝑖𝑚 𝛿𝑗𝑘 𝜀𝑘𝑚

08/07/2017 Dr. Luis Mosquera L. 248


this reduces to
𝑡𝑖𝑗 = 𝜆𝛿𝑖𝑗 𝜀𝑘𝑘 + 2𝜇 𝜀𝑖𝑗
which is Hooke’s law for isotropic elastic behavior

λ and μ, are known as the Lamé constants.

Example:

Show that for an isotropic linear elastic solid the principal axes of the stress and
strain tensors coincide, and develop an expression for the relationship among their
principal values.
Solution

Let 𝑛 𝑞 , (q = 1, 2, 3) be unit normals in the principal directions of εij, and associated


with these normals the corresponding principal values ε(q).

08/07/2017 Dr. Luis Mosquera L. 249


(𝑞) (𝑞) (𝑞) (𝑞)
𝑡𝑖𝑗 𝑛𝑗 = 𝜆𝛿𝑖𝑗 𝜀𝑘𝑘 + 2𝜇 𝜀𝑖𝑗 𝑛𝑗 = 𝜆𝑛𝑖 𝜀𝑘𝑘 + 2𝜇 𝜀𝑖𝑗 𝑛𝑗

but (𝑞) (𝑞)


𝜀𝑖𝑗 𝑛𝑗 = 𝜀(𝑞) 𝛿𝑖𝑗 𝑛𝑗

So that
(𝑞) (𝑞) (𝑞)
𝑡𝑖𝑗 𝑛𝑗 = 𝜆𝜀𝑘𝑘 + 2𝜇 𝜀(𝑞) 𝑛𝑖 = 𝜆𝜀𝑘𝑘 + 2𝜇 𝜀(𝑞) 𝑛𝑖

and because 𝜀𝑘𝑘 = 𝜀 1 + 𝜀 2 + 𝜀 3 is the first invariant of strain, it is constant for all
(𝑞)
𝑛𝑖 so that
(𝑞) (𝑞)
𝑡𝑖𝑗 𝑛𝑗 = 𝜆 𝜀(1) + 𝜀(2) + 𝜀(3) + 2𝜇𝜀(𝑞) 𝑛𝑗

(𝑞)
This indicates that 𝑛𝑖 are principal directions of stress also, with principal stress
values
𝑡(𝑞) = 𝜆 𝜀(1) + 𝜀(2) + 𝜀(3) + 2𝜇𝜀(𝑞)

08/07/2017 Dr. Luis Mosquera L. 250


Expressing the strain components in terms of the stresses

𝑡𝑖𝑖 = 3𝜆𝜀𝑘𝑘 + 2𝜇𝜀𝑖𝑖 = 3𝜆 + 2𝜇 𝜀𝑖𝑖

Thus, we obtain the inverse form of the isotropic constitutive equation,

1 𝜆
𝜀𝑖𝑗 = 𝑡 − 𝛿 𝑡
2𝜇 𝑖𝑗 3𝜆 + 2𝜇 𝑖𝑗 𝑘𝑘

rearrangement of this equation, we may write

𝜆+𝜇 𝜆 𝜆
𝜀𝑖𝑗 = 1+ 𝑡𝑖𝑗 − 𝛿 𝑡
𝜇(3𝜆 + 2𝜇) 2 𝜆+𝜇 2 𝜆 + 𝜇 𝑖𝑗 𝑘𝑘

from which if we define


𝜆
𝜇(3𝜆 + 2𝜇) and 𝜈=
𝐸= 2(𝜆 + 𝜇)
𝜆+𝜇

08/07/2017 Dr. Luis Mosquera L. 251


we obtain the following form of Hooke’s law for isotropic behavior in terms of the
engineering constants E and v,

1
𝜀𝑖𝑗 = 1 + 𝜈 𝑇𝑖𝑗 − 𝜈𝛿𝑖𝑗 𝑇𝑘𝑘
𝐸
𝐸
𝑇𝑖𝑗 = 𝜈𝛿𝑖𝑗 𝜀𝑘𝑘 + 1 − 2𝜈 𝜀𝑖𝑗
1 + 𝜈 1 − 2𝜈

E is called Young’s modulus, and v is known as Poisson’s ratio.


The shear modulus, or modulus of rigidity (G) and the bulk modulus (K) are
defined as:

𝐸 𝐸
𝐺= =𝜇 𝐾=
2(1 + 𝜈) 3(1 − 2𝜈)

08/07/2017 Dr. Luis Mosquera L. 252


The physical interpretations of the constants E, v, G, and K introduced above can
be determined from a consideration of the special states of stress displayed in
figure:

In the case of a uniaxial state of stress (tension or compression), say in the x1


direction with σ11 = ±σ0 and all other stress components are zero.

𝑡11 ±𝜎0
𝜀11 = = 𝑓𝑜𝑟 (𝑖 = 𝑗 = 1)
𝐸 𝐸
∓𝜈𝜎0
𝜀22 = −𝜈𝜀11 = 𝑓𝑜𝑟 (𝑖 = 𝑗 = 2)
𝐸
∓𝜈𝜎0
𝜀33 = −𝜈𝜀11 = 𝑓𝑜𝑟 (𝑖 = 𝑗 = 3)
08/07/2017 Dr. Luis𝐸Mosquera L. 253
ν is seen to be the ratio of the unit lateral contraction to unit longitudinal
extension for tension, and vice versa for compression.

For the simple shear case shown we, say, σ12 = τ0 , all other stresses zero,

1+𝜈 𝜏0
𝜀12 = 𝑡12 =
𝐸 2𝐺
or for engineering strains,
𝑡12 𝜏0
𝛾12 = =
𝐺 𝐺

Finally, for the case of uniform triaxial tension (or hydrostatic compression), we
take 𝒕𝒊𝒋 = ±𝒑𝜹𝒊𝒋 with p>0. For this,

1 − 2𝜈 ±3 1 − 2𝜈 ±𝑝
𝜀𝑖𝑖 = 𝑡𝑖𝑖 = 𝑝=
𝐸 𝐸 𝑘
by which we infer that the bulk modulus K relates the pressure p to the volume
change given by the cubical dilation εii

08/07/2017 Dr. Luis Mosquera L. 254


By use of the constants G and K, Hooke’s law may be expressed in terms of the
spherical and deviator components of the stress and strain tensors. Thus, the pair
of equations

𝑆𝑖𝑗 = 2𝐺𝜂𝑖𝑗

𝑡𝑖𝑖 = 3𝑘𝜀𝑖𝑖
may be shown to be equivalent to

1
𝜀𝑖𝑗 = 1 + 𝜈 𝑇𝑖𝑗 − 𝜈𝛿𝑖𝑗 𝑇𝑘𝑘
𝐸

08/07/2017 Dr. Luis Mosquera L. 255


Elastic Symmetry; Hooke’s Law for
Anisotropic Media

Hooke’s law for isotropic behavior was established on the basis of C being a
fourth-order isotropic tensor. The same result may be achieved from the concepts
of elastic symmetry.

Defined equivalent elastic directions as those specified by Cartesian axes Ox1x2x3 and
Ox’1x’2x’3 at a point such that the elastic constants Cαβ are unchanged by a
transformation between the two sets of axes.

If the transformation represents a rotation about an axis, we say the material has
axial elastic symmetry with respect to that axis.

the x3 axis in figure a is said to be of


order N where N = 2π/θ

08/07/2017 Dr. Luis Mosquera L. 256


If the transformation is a reflection of the axes with respect to some plane, we
say the material has a plane of elastic symmetry.

It is also noteworthy that a point of elastic symmetry would imply isotropic


behavior, since the elastic constants would remain unchanged for any two sets of
Cartesian axes at the point.

08/07/2017 Dr. Luis Mosquera L. 257


Let us consider the consequences of the x1x2 plane being a plane of elastic symmetry
The transformation matrix for this is clearly

1 0 0
𝑎𝑖𝑗 = 0 1 0
0 0 −1

so that in the single subscript notation for stress and strain components the
transformations in matrix form are

𝑡1′ 𝑡6′ 𝑡5′ 1 0 0 𝑡1 𝑡6 𝑡5 1 0 0


𝑡6′ 𝑡2′ ′
𝑡4 = 0 1 0 𝑡6 𝑡2 𝑡4 0 1 0
𝑡5′ 𝑡4′ 𝑡3′ 0 0 −1 𝑡5 𝑡4 𝑡3 0 0 −1
𝑡1 𝑡6 −𝑡5
= 𝑡6 𝑡2 −𝑡4
−𝑡5 −𝑡4 𝑡3

08/07/2017 Dr. Luis Mosquera L. 258


and

1 ′ 1 ′ 1 1
𝜖1′ 𝜖6 𝜖5 𝜖1 𝜖6 𝜖5
2 2 2 2
1 ′ 1 ′ 1 0 0 1 1 1 0 0
𝜖 𝜖2′ 𝜖4 = 0 1 0 𝜖6 𝜖2 𝜖4 0 1 0
2 6 2 0 0 −1 2 2 0 0 −1
1 ′ 1 ′ 1 1
𝜖 𝜖 𝜖3′ 𝜖 𝜖 𝜖3
2 5 2 4 2 5 2 4
1 1
𝜖1 𝜖6 − 𝜖5
2 2
1 1
= 𝜖6 𝜖2 − 𝜖4
2 2
1 1
− 𝜖5 − 𝜖4 𝜖3
2 2

08/07/2017 Dr. Luis Mosquera L. 259


Therefore, assuming all 36 constants in equation

𝑡𝛼 = 𝒞𝛼𝛽 𝜖𝛽

are distinct, we note that for axes Ox1x2x3

𝑡1 = 𝒞11 𝜖1 + 𝒞12 𝜖2 + 𝒞13 𝜖3 + 𝒞14 𝜖4 + 𝒞15 𝜖5 + 𝒞16 𝜖6

whereas for axes Ox’1x’2x’3, under the condition that x1x2 is a plane of symmetry
such that the Cαβ are unchanged in this system, we have

𝑡′1 = 𝒞11 𝜖′1 + 𝒞12 𝜖′2 + 𝒞13 𝜖′3 + 𝒞14 𝜖′4 + 𝒞15 𝜖′5 + 𝒞16 𝜖′6

But as, 𝑡𝛼′ = 𝑡𝛼 𝛼 = 1,2,3,6 𝑎𝑛𝑑 𝑡𝛼′ = −𝑡𝛼 (𝛼 = 4,5)

Likewise, 𝜖𝛼′ = 𝜖𝛼 𝛼 = 1,2,3,6 𝑎𝑛𝑑 𝜖𝛼′ = −𝜖𝛼 (𝛼 = 4,5)

08/07/2017 Dr. Luis Mosquera L. 260


So that,

𝑡′1 = 𝑡1 = 𝒞11 𝜖1 + 𝒞12 𝜖2 + 𝒞13 𝜖3 − 𝒞14 𝜖4 − 𝒞15 𝜖5 + 𝒞16 𝜖6

Comparing these expressions for t1, we must have 𝒞14 = 𝒞15 = 0

Following the same procedure, the additional elastic constants C24, C25, C34, C35,
C41, C42, C43, C46, C51, C52, C53, C56, C64, and C65 must also be zero. for the x1x2 plane
to be one of elastic symmetry
Accordingly, the elastic constant matrix for this case has the form
𝒞11 𝒞12 𝒞13 0 0 𝒞16
𝒞21 𝒞22 𝒞23 0 0 𝒞26 for the x1x2 plane to be one of elastic
𝒞𝛼𝛽
𝒞
= 31
𝒞32 𝒞33 0 0 𝒞36 symmetry
0 0 0 𝒞44 𝒞45 0
0 0 0 𝒞54 𝒞55 0
𝒞61 𝒞62 𝒞63 0 0 𝒞66

And the original 36 constants are reduced to 20.


Also, if a strain energy functions exists, Cαβ = Cβα and these 20 nonzero constants
would be further reduced to 13.
08/07/2017 Dr. Luis Mosquera L. 261
If the x2x3 plane is also one of elastic symmetry at the same time as the x1x2 plane
at a point and we repeat the procedure outlined above, we find that C16, C26, C36, C45,
C54, C61, C62, and C63 must also be zero, and the C matrix is further reduced to

𝒞11 𝒞12 𝒞13 0 0 0


𝒞21 𝒞22 𝒞23 0 0 0
0 0 0
𝒞𝛼𝛽 = 𝒞31 𝒞32 𝒞33
0 0 0 𝒞44 0 0
0 0 0 0 𝒞55 0
0 0 0 0 0 𝒞66

having 12 nonzero coefficients, or 9 if a strain energy function exists.

Interestingly enough, if x1x3 is also a plane of elastic symmetry along with the two
considered above, no further reduction in the Cαβ matrix occurs.

A material possessing three mutually perpendicular planes of elastic symmetry is


called an orthotropic material, and its elastic constants matrix is that given.

08/07/2017 Dr. Luis Mosquera L. 262


The reduction of the orthotropic elastic matrix to that of the isotropic matrix may
be completed by successive consideration of the three axes of elastic symmetry
shown in Figure

08/07/2017 Dr. Luis Mosquera L. 263


08/07/2017 Dr. Luis Mosquera L. 264
08/07/2017 Dr. Luis Mosquera L. 265
By the rotation of 90° about the x1 axis (Figure a), we find that C12 = C13, C21 = C31, C22
= C33, C23 = C32, and C55 = C66. For the rotation of 90° about the x3 axis (Figure b), we
see that C12 = C21, C11 = C22, C12 = C23, C31 = C32, and C44 = C55. Finally, by a rotation of
45° about the x3 axis (Figure c), we obtain 2C44 = C11 – C12. Therefore, by setting C44 =
μ and C12 = λ, we may write the elastic coefficient matrix for isotropic behavior as

𝜆 + 2𝜇 𝜆 𝜆 0 0 0
𝜆 𝜆 + 2𝜇 𝜆 0 0 0
𝜆 𝜆 𝜆 + 2𝜇 0 0 0
𝒞𝛼𝛽 =
0 0 0 𝜇 0 0 for isotropic behavior
0 0 0 0 𝜇 0
0 0 0 0 0 𝜇

08/07/2017 Dr. Luis Mosquera L. 266


this matrix may be expressed in terms of the engineering constants E and v, as

for isotropic behavior

𝑡1 0 0 0 𝜖1
1−𝜈 𝜈 𝜈
𝑡2 0 0 0 𝜖2
𝜈 1−𝜈 𝜈 0 0 0
𝑡3 𝐸 𝜈 𝜈 1−𝜈 𝜖3
= 1 0 0
𝑡4 1 + 𝜈 1 − 2𝜈 0 0 0 1 − 2𝜈 𝜖4
0 2 1 0 𝜖5
𝑡5 0 0 1 − 2𝜈 1
0 0 0 0 2 𝜖6
𝑡6 1 − 2𝜈
0 0 2

08/07/2017 Dr. Luis Mosquera L. 267


Isotropic Elastostatics and Elastodynamics,
Superposition Principle

The formulation and solution of the basic problems of linear elasticity comprise the
subjects we call elastostatics and elastodynamics. Elastostatics is restricted to
those situations in which inertia forces may be neglected. In both elastostatics
and elastodynamics, certain field equations have to be satisfied at all interior
points of the elastic body under consideration, and at the same time the field
variables must satisfy specific conditions on the boundary. In the case of
elastodynamics problems, initial conditions on velocities and displacements must
also be satisfied.

08/07/2017 Dr. Luis Mosquera L. 268


We begin with elastostatics for which the appropriate field equations are:

(a) Equilibrium equations

𝑡𝑗𝑖 ,𝑗 + 𝜌𝑏𝑖 = 0

(b) Strain-displacement relation

2𝜀𝑖𝑗 = 𝑢𝑖,𝑗 + 𝑢𝑗 ,𝑖
(c) Hooke’s law
𝑡𝑖𝑗 = 𝜆𝛿𝑖𝑗 𝜀𝑘𝑘 + 2𝜇𝜀𝑖𝑗
or 1
𝜀𝑖𝑗 = 1 + 𝜈 𝑡𝑖𝑗 − 𝜈𝛿𝑖𝑗 𝑡𝑘𝑘
𝐸
It is usually assumed that the body forces bi are known so that the solution we
seek from the fifteen equations listed here is for the six stresses tij, the six strains
εij, and the three displacements ui.

08/07/2017 Dr. Luis Mosquera L. 269


The conditions to be satisfied on the boundary surface S will appear in one of
the following statements:
1. Displacements prescribed everywhere,

𝑢𝑖 = 𝑢𝑖∗ 𝑥 𝑜𝑛 𝑆

2. Tractions prescribed everywhere,


𝑛 ∗ 𝑛
𝑡𝑖 = 𝑡𝑖 𝑜𝑛 𝑆

3. Displacements prescribed on portion S1 of S,

𝑢𝑖 = 𝑢𝑖∗ 𝑥 𝑜𝑛 𝑆1

with tractions prescribed on the remainder S2,

𝑛 ∗ 𝑛
𝑡𝑖 = 𝑡𝑖 𝑜𝑛 𝑆2

08/07/2017 Dr. Luis Mosquera L. 270


A most important feature of the field equations is that they are linear in the
unknowns. Consequently, if , 𝑡𝑖𝑗1 , and 𝜀𝑖𝑗1 are a solution for body forces 1 𝑏𝑖∗
1 ∗(𝑛 )
And surface tractions 𝑡𝑖 , whereas 𝑡𝑖𝑗(2) , 𝜀𝑖𝑗(2) , 𝑎𝑛𝑑 𝑢𝑖(2) are a solution for body
forces 2 𝑏𝑖∗ and surface tractions 2 ∗ 𝑛
𝑡𝑖 , then

1 2
𝑡𝑖𝑗 = 𝑡𝑖𝑗 + 𝑡𝑖𝑗

1 2
𝜀𝑖𝑗 = 𝜀𝑖𝑗 + 𝜀𝑖𝑗

1 2
𝑢𝑖 = 𝑢𝑖 + 𝑢𝑖

offer a solution for the situation where

𝑏𝑖 = 1𝑏𝑖∗ + 2𝑏𝑖∗ This is a statement of the principle of


superposition
𝑛 ∗ 𝑛 ∗ 𝑛
𝑡𝑖 = 1𝑡𝑖 + 2𝑡𝑖

08/07/2017 Dr. Luis Mosquera L. 271


For those problems in which the boundary conditions are given in terms of
displacements by 𝑢𝑖 = 𝑢𝑖∗ 𝑥 𝑜𝑛 𝑆 , it is convenient for us to eliminate the stress
and strain unknowns from the field equations so as to state the problem
solely in terms of the unknown displacement components. Thus,

𝑡𝑗𝑖 ,𝑗 + 𝜌𝑏𝑖 = 0 𝜇𝑢𝑖,𝑗𝑗 + 𝜆 + 𝜇 𝑢𝑗 ,𝑗𝑖 + 𝜌𝑏𝑖 = 0

which are known as the Navier equations.

When the boundary conditions are given in terms of surface tractions


the equations of compatibility for infinitesimal strains may be combined with
Hooke’s law and the equilibrium equations to arrive at the equations,
1 𝜈
𝑡𝑖𝑗 ,𝑘𝑘 + 𝑡𝑘𝑘 ,𝑖𝑗 + 𝜌 𝑏𝑖,𝑗 + 𝑏𝑗 ,𝑖 + 𝛿𝑖𝑗 𝜌𝑏𝑘,𝑘 = 0
1+𝜈 1−𝜈

which are known as the Beltrami-Michell stress equations of compatibility.


08/07/2017 Dr. Luis Mosquera L. 272
In elastodynamics, the equilibrium equations must be replaced by the equations of
motion in the system of basic field equations. So that a solution for the displacement
field appears in the form

𝑢𝑖 = 𝑢𝑖∗ 𝑥, 𝑡 𝑜𝑛 𝑆
𝑛 ∗ 𝑛
𝑡𝑖 = 𝑡𝑖 (𝑥, 𝑡) 𝑜𝑛 𝑆

with initial conditions


𝑢𝑖 = 𝑢𝑖∗ 𝑥, 0
and
𝑢𝑖 = 𝑢𝑖∗ 𝑥, 0

the governing equations for displacements in elastodynamics theory are

𝜇𝑢𝑖,𝑗𝑗 + 𝜆 + 𝜇 𝑢𝑗 ,𝑗𝑖 + 𝜌𝑏𝑖 = 𝜌𝑢𝑖

which are the Navier’s equations.

08/07/2017 Dr. Luis Mosquera L. 273


08/07/2017 Dr. Luis Mosquera L. 274
SAINT VENANT PROBLEM

Solving an elasticity problem with the boundary conditions is a daunting task. In


the 1800s Saint-Venant studied long beams that were loaded in a variety of ways:
extension, torsion, pure bending and flexure. Rather than satisfy exact boundary
conditions on the lateral surface and the ends, Saint Venant solved the problems
by considering relaxed boundary conditions on the beam ends. If one considers
statically equivalent force and moment systems, the solution sufficiently far
away from the application of the load will be the same. This is known as the
Saint-Venant Principle.

08/07/2017 Dr. Luis Mosquera L. 275


To demonstrate Saint Venant’s solution considerar a beam as shown in the figure:

The lateral surface of the cylinder βL, will be stress free, This condition can be written
as

𝑛
𝑡𝑖 = 𝑡𝑖1 𝑛1 + 𝑡𝑖2 𝑛2 = 0 𝑜𝑛 ℬ𝐿

08/07/2017 Dr. Luis Mosquera L. 276


For the different problems (extensión, torsion, pure bending and flexure) the ends,
C1 and C2, will have relaxed boundary conditions. The sum of the stress over the
end’s area will give a force vector Ti

𝑡3𝑖 𝑑𝑆 = 𝑇𝑖
𝒞1

Bending moments on the ends are expressed in terms of the axial stress t33 as:

𝑡33 𝑥2 𝑑𝑆 = 𝑀1 , 𝑡33 𝑥1 𝑑𝑆 = −𝑀2


𝒞1 𝒞1

Finally, the torque on the beam’s end may be written as

𝑡33 𝑥2 − 𝑡31 𝑥2 𝑑𝑆 = 𝑀3 = 𝑀𝑡
𝒞1

Similar equations hold for end C2


08/07/2017 Dr. Luis Mosquera L. 277
A reasonable assumption was made regarding the stress components, strain or
displacements. From this assumption the equilibrium and compatibility equations
were shown to be satisfied. Boundary conditions, exact and relaxed, were
checked. The different problems are solved by setting the components of the end
force Ti and moment Mi to different values:

𝐼 𝑇1 = 𝑇2 = 0, 𝑇3 = 𝑇, 𝑀𝑖 = 0 𝑃𝑢𝑟𝑒 𝑒𝑥𝑡𝑒𝑛𝑠𝑖𝑜𝑛

𝐼𝐼 𝑇𝑖 = 0, 𝑀1 = 𝑀2 = 0, 𝑀3 = 𝑀𝑡 𝑃𝑢𝑟𝑒 𝑡𝑜𝑟𝑠𝑖𝑜𝑛

𝐼𝐼𝐼 𝑇𝑖 = 0, 𝑀1 = 𝑀3 = 0, 𝑀2 = 𝑀 𝑃𝑢𝑟𝑒 𝑏𝑒𝑛𝑑𝑖𝑛𝑔

𝐼𝑉 𝑇1 = 𝑇, 𝑇2 = 𝑇3 = 0, 𝑀𝑖 = 0 𝑃𝑢𝑟𝑒 𝑓𝑙𝑒𝑥𝑢𝑟𝑒

More complex elasticity problems can be obtained by superposing the basic solutions.

08/07/2017 Dr. Luis Mosquera L. 278


Extension

For the case of pure extension the nonzero stress components are assumed to be
𝑇
𝑡33 =
𝐴

Where the area is 𝑑𝑆


𝒞

And all other components are zero. These stress components clearly satisfy the
equilibrium conditions without body forces. Also, the relaxed boundary conditions
are satisfied.The strains can be found to be

𝑇 𝜈𝑇
ℰ33 = , ℰ11 = ℰ22 = −
𝐸𝐴 𝐸𝐴

08/07/2017 Dr. Luis Mosquera L. 279


With all other strain components being zero. The compatibility condition is satisfied
since the strains are constant, and using the strain-displacement relationship, the
displacements can be found to be,

𝑇 𝜈𝑇 𝜈𝑇
𝑢3 = 𝑥 , 𝑢1 = − 𝑥 , 𝑢2 = − 𝑥
𝐸𝐴 3 𝐸𝐴 1 𝐸𝐴 2

08/07/2017 Dr. Luis Mosquera L. 280


Torsion

We begin with a brief review of the solution to the case of a shaft having a
constant circular cross section when subjected to equilibrating end couples, Mt
as shown in the figure (a). Let the end face at x3 = 0 be fixed while the face at
x3 = L is allowed to rotate about the axis of the shaft.

It is assumed that plane sections perpendicular


to the axis remain plane under the twisting,
and that each rotates through an angle
proportional θ x3 to its distance from the fixed
end.

08/07/2017 Dr. Luis Mosquera L. 281


A point in the cross section at
coordinate x3 will rotate an angle of θ x3.
Each point, say point P, in the cross
section travels a distance θ x3R which is
proportional to the distance R from the
x3 axis as shown in figure b

08/07/2017 Dr. Luis Mosquera L. 282


The distance squared to point P is the square of the x1 and x2 coordinates (figure c).
Using this distance, it is possible to define the cosine and sine of angle β.

𝑐𝑜𝑠 90 − 𝛽 = 𝑥2 /𝑅

and

𝑠𝑖𝑛 90 − 𝛽 = 𝑥1 /𝑅
(c)

Thus,
𝑢1 = −𝜃𝑥3 𝑅𝑐𝑜𝑠 90 − 𝛽 = −𝜃𝑥3 𝑥2

𝑢2 = 𝜃𝑥3 𝑥1 , 𝑢3 = 0

where θ is the angle of twist per unit length of the shaft.


08/07/2017 Dr. Luis Mosquera L. 283
Recall that
2𝜀𝑖𝑗 = 𝑢𝑖,𝑗 + 𝑢𝑗 ,𝑖

1
0 0 − 𝜃𝑥2
2
1
𝜀= 0 0 𝜃𝑥
2 1
1 1
− 𝜃𝑥2 𝜃𝑥 0
2 2 1

and since, 𝜇 ≡ 𝐺 , we obtain the stress components

𝑡23 = 𝐺𝜃𝑥1 , 𝑡13 = −𝐺𝜃𝑥2 , 𝑡11 = 𝑡22 = 𝑡33 = 𝑡12 = 0

Because these stress components, as well as the strains from which they were
derived, are either linear functions of the coordinates or zero, the compatibility
equations are satisfied. Likewise, for zero body forces, the equilibrium equations
are clearly satisfied.
08/07/2017 Dr. Luis Mosquera L. 284
The lateral surface of the shaft is stress free. To verify this, consider the stress
components in the direction of the normal at a point on the cross-section
perimeter designated in figure b. Thus, at a radius R = a,

x2
𝑥1
(x1 ,x2) 𝑛1 =
𝑎
a
𝑥2
x1 𝑛2 =
𝑎

𝑛
𝑡𝑖 = 𝑡𝑖1 𝑛1 + 𝑡𝑖2 𝑛2 = 0 𝑜𝑛 ℬ𝐿

𝑥1 𝑥2 𝐺𝜃
𝑡13 + 𝑡23 = −𝑥2 𝑥1 + 𝑥1 𝑥2 = 0
𝑎 𝑎 𝑎

08/07/2017 Dr. Luis Mosquera L. 285


At the same time, the total shearing stress at any point of the cross section
is the resultant

2 2
𝜏= 𝑡13 + 𝑡23 = 𝐺𝜃 𝑥12 + 𝑥22 = 𝐺𝜃𝑅

which indicates that the shear is proportional to the radius at the point, and
perpendicular to that radius. By summing the moments of the shear forces on
either end face of the shaft, we find that

𝑀𝑡 = 𝑥1 𝑡23 + 𝑥2 𝑡13 𝑑𝑥1 𝑑𝑥2 = 𝐺𝜃 𝑅 2 𝑑𝑥1 𝑑𝑥2 = 𝐺𝜃𝐼𝑃

where Ip is the polar moment of inertia of the cross section.

08/07/2017 Dr. Luis Mosquera L. 286


For a prismatic shaft of any cross section other than circular, plane sections do not
remain plane under twisting, and warping will occur. For such cases we must expressing
the displacements in the form

𝑢1 = −𝜃𝑥2 𝑥3 , 𝑢2 = 𝜃𝑥1 𝑥3 , 𝑢3 = 𝜃𝜓 𝑥1 , 𝑥2

Where 𝜓 𝑥1 , 𝑥2 is called the warping function.

𝜃 𝜕𝜓
0 0 − 𝑥2
2 𝜕𝑥1
𝜃 𝜕𝜓
𝜀= 0 0 + 𝑥1
2 𝜕𝑥2
𝜃 𝜕𝜓 𝜃 𝜕𝜓
− 𝑥2 + 𝑥1 0
2 𝜕𝑥1 2 𝜕𝑥2

08/07/2017 Dr. Luis Mosquera L. 287


𝑡𝑖𝑗 = 𝜆𝛿𝑖𝑗 𝜀𝑘𝑘 + 2𝜇𝜀𝑖𝑗

𝜕𝜓
0 0 𝜇𝜃 − 𝑥2
𝜕𝑥1
𝜕𝜓
𝑡= 0 0 𝜇𝜃 + 𝑥1
𝜕𝑥2
𝜕𝜓 𝜕𝜓
𝜇𝜃 − 𝑥2 𝜇𝜃 + 𝑥1 0
𝜕𝑥1 𝜕𝑥2

𝑡23 = 𝐺𝜃(𝜓,2 + 𝑥1 ), 𝑡13 = 𝐺𝜃(𝜓,1 − 𝑥2 ), 𝑡11 = 𝑡22 = 𝑡33 = 𝑡12 = 0

It is clear from these stress components that there are no normal stresses
between the longitudinal elements of the shaft.

08/07/2017 Dr. Luis Mosquera L. 288


0
𝑡𝑗𝑖 ,𝑗 + 𝜌𝑏𝑖 = 0

𝑡1𝑖,1 + 𝑡2𝑖,2 + 𝑡3𝑖,3 = 0

The first two of the equilibrium equations, are satisfied identically in the absence
of body forces, and substitution into the third equilibrium equation yields

𝑡13,1 + 𝑡23,2 + 𝑡33,3 = 0

𝜕2 𝜓 𝜕2 𝜓
𝜇𝜃 2 + 𝜇𝜃 2 = 0 = 𝐺𝜃 𝜓,11 + 𝜓,22 = 0
𝜕𝑥1 𝜕𝑥2

which indicates that ψ must be harmonic,

∇2 𝜓 = 0
on the cross section of the shaft.

08/07/2017 Dr. Luis Mosquera L. 289


Boundary conditions on the surfaces of the shaft must also be satisfied. On the
lateral surface which is stress free, the following conditions must prevail

𝑡. 𝑛 = 0

𝑡11 𝑛1 + 𝑡12 𝑛2 = 0, 𝑡21 𝑛1 + 𝑡22 𝑛2 = 0, 𝑡31 𝑛1 + 𝑡32 𝑛2 = 0,

here n3 = 0. The first two of these equations are satisfied identically while the
third requires

𝐺𝜃 𝜓,1 − 𝑥2 𝑛1 + 𝐺𝜃 𝜓,2 + 𝑥1 𝑛2 = 0

which reduces to
𝑑𝜓
𝜓,1 𝑛1 + 𝜓,2 𝑛2 = = 𝑥2 𝑛1 − 𝑥1 𝑛2
𝑑𝑛

08/07/2017 Dr. Luis Mosquera L. 290


We note further that in order for all cross sections to be force free, that is, in simple
shear over those cross sections

𝑡13 𝑑𝑥1 𝑑𝑥2 = 𝑡23 𝑑𝑥1 𝑑𝑥2 = 𝑡33 𝑑𝑥1 𝑑𝑥2 = 0

On the end faces of the shaft, x3 = 0 or x3 = L, the following conditions must be


satisfied

𝑥2 𝑡33 𝑑𝑥1 𝑑𝑥2 = 𝑥1 𝑡33 𝑑𝑥1 𝑑𝑥2 = 0, 𝑥1 𝑡23 − 𝑥2 𝑡13 𝑑𝑥1 𝑑𝑥2 = 𝑀𝑡

Again, since σ33 = 0, the first two of these are trivial. The third leads to

𝑀𝑡 = 𝐺𝜃 𝑥12 + 𝑥22 + 𝑥1 𝜓,2 − 𝑥2 𝜓,1 𝑑𝑥1 𝑑𝑥2

08/07/2017 Dr. Luis Mosquera L. 291


Defining the torsional rigidity as

𝐾=𝐺 𝑥12 + 𝑥22 + 𝑥1 𝜓,2 − 𝑥2 𝜓,1 𝑑𝑥1 𝑑𝑥2

which can be evaluated once 𝜓 𝑥1 , 𝑥2 is known. We express the angle of twist as

𝑀𝑡
𝜃=
𝐾

08/07/2017 Dr. Luis Mosquera L. 292


A second approach to the general torsion problem rests upon the introduction of a
torsion stress function, designated here by Φ and defined so that the non-zero
stresses are related to it by the definitions

𝜕Φ 𝜕Φ
𝑡13 = , 𝑡23 =−
𝜕𝑥2 𝜕𝑥1

Thus, 𝜕Φ 𝜕Φ
= 𝐺𝜃 𝜓,1 − 𝑥2 , = −𝐺𝜃 𝜓,2 + 𝑥1
𝜕𝑥2 𝜕𝑥1

From this pair of equations we obtain

∇2 Φ = −2𝐺𝜃

08/07/2017 Dr. Luis Mosquera L. 293


As already noted, the lateral surface of the shaft parallel to the axis must remain
stress free, that is, the equation 𝑡31 𝑛1 + 𝑡32 𝑛2 = 0 must be satisfied

However, it is advantageous to write this condition in terms of the unit vector 𝑠


along the boundary rather than unit normal 𝑛 as shown in figure:

𝑑𝑥2 𝑑𝑥1
𝑡31 𝑛1 + 𝑡32 𝑛2 = 0 𝑡13 − 𝑡23 =0
𝑑𝑠 𝑑𝑠

08/07/2017 Dr. Luis Mosquera L. 294


which in terms of Φ becomes
𝜕Φ 𝑑𝑥1 𝜕Φ 𝑑𝑥2 𝑑Φ
+ = =0
𝜕𝑥1 𝑑𝑠 𝜕𝑥2 𝑑𝑠 𝑑𝑠

Thus, Φ is a constant along the perimeter of the cross section and will be assigned
the value of zero here.

In order for all cross sections to be force free,

𝑡13 𝑑𝑥1 𝑑𝑥2 = 𝑡23 𝑑𝑥1 𝑑𝑥2 = 𝑡33 𝑑𝑥1 𝑑𝑥2 = 0

In terms of Φ

∂Φ ∂Φ
𝑑𝑥 𝑑𝑥 = 𝑑𝑥 𝑑𝑥1 = Φ 𝑏𝑎 𝑑𝑥1 = 0
𝜕𝑥2 1 2 𝜕𝑥2 2

since Φ is constant on the perimeter. Likewise, by the same reasoning, the second
of equations is satisfied, while the third is satisfied since τ33 = 0.
08/07/2017 Dr. Luis Mosquera L. 295
On the end faces of the shaft, x3 = 0 or x3 = L, the following conditions must be
satisfied

𝑥2 𝑡33 𝑑𝑥1 𝑑𝑥2 = 𝑥1 𝑡33 𝑑𝑥1 𝑑𝑥2 = 0, 𝑥1 𝑡23 − 𝑥2 𝑡13 𝑑𝑥1 𝑑𝑥2 = 𝑀𝑡

∂Φ ∂Φ
−𝑥1 − 𝑥2 𝑑𝑥1 𝑑𝑥2 = 𝑀𝑡
𝜕𝑥1 𝜕𝑥2

Integrating here by parts and using the fact that Φ is assumed zero on the
perimeter C yields

𝑀𝑡 = 2 Φ 𝑑𝑥1 𝑑𝑥2

Thus, the solution by this approach consists of determining the stress function
Φ which is zero on the cross-section perimeter, and satisfies

∇2 Φ = −2𝐺𝜃
08/07/2017 Dr. Luis Mosquera L. 296
Example
Determine the stresses and the angle of twist for a solid elliptical shaft of the
dimensions shown when subjected to end couples Mt .

08/07/2017 Dr. Luis Mosquera L. 297


Solution
The equation of this ellipse is given by

𝑥12 𝑥22
2
+ 2=1
𝑎 𝑏
Therefore, take the stress function Φ in the form

𝑥12 𝑥22
Φ=𝜆 2+ 2 −1
𝑎 𝑏

where λ is a constant. Thus, Φ is zero on the cross-section perimeter. From the


equation ∇2 Φ = −2𝐺𝜃

1 1
2𝜆 + = −2𝐺𝜃
𝑎2 𝑏 2

so that
𝑎2 𝑏 2 𝐺𝜃
𝜆=− 2
𝑎 + 𝑏2
08/07/2017 Dr. Luis Mosquera L. 298
Now from equation
𝑀𝑡 = 2 Φ 𝑑𝑥1 𝑑𝑥2

Then,

2𝑎2 𝑏 2 𝐺𝜃 𝑥12 𝑥22


𝑀𝑡 = − 2 + − 1 𝑑𝑥1 𝑑𝑥2
𝑎 + 𝑏2 𝑎2 𝑏 2

Noting that
1
𝑥12 𝑑𝑥1 𝑑𝑥2 = 𝐼𝑥 2 = 𝜋𝑏𝑎3
4
1
𝑥22 𝑑𝑥1 𝑑𝑥2 = 𝐼𝑥 1 = 𝜋𝑎𝑏 3
4

𝑑𝑥1 𝑑𝑥2 = 𝜋𝑎𝑏

08/07/2017 Dr. Luis Mosquera L. 299


we may solve for Mt which is

𝜋𝑎3 𝑏 3 𝐺𝜃
𝑀𝑡 = 2
𝑎 + 𝑏2
Therefore

𝑀𝑡 𝑥12 𝑥22
Φ=− + −1
𝜋𝑎𝑏 𝑎2 𝑏 2

𝜕Φ 2𝑀𝑡 𝜕Φ 2𝑀𝑡
𝑡13 = =− 𝑥 , 𝑡23 =− = 𝑥
𝜕𝑥2 𝜋𝑎𝑏 3 2 𝜕𝑥1 𝜋𝑎3 𝑏 1

The maximum stress occurs at the ends of the minor axis, and equals
2𝑀𝑡
𝜏𝑚𝑎𝑥 =± ;(X2=b)
𝜋𝑎𝑏 2

08/07/2017 Dr. Luis Mosquera L. 300


The torsional rigidity is calculated to be

𝑀𝑡 𝜋𝑎3 𝑏 3 𝐺 𝐺𝐴4
𝑘= = 2 =
𝜃 𝑎 + 𝑏 2 4𝜋 2 𝐼𝑃

for a = b (circular cross section), the resultant stress at any point is

2 2
𝑀𝑡
𝜏= 𝑡13 + 𝑡23 = 𝑟
𝐼𝑃

08/07/2017 Dr. Luis Mosquera L. 301


Pure Bending
Consider the beam to be subject to end conditions

𝑇𝑖 = 0, 𝑀1 = 𝑀3 = 0, 𝑀2 = 𝑀

And assume the stresses to be


𝑀𝑥1
𝑡33 =−
𝐼

with all other components zero.

The compatibility equations are satisfied since the nonzero stresses are linear in the
coordinates. The equilibrium equations tji,j = 0 only have one equation which is not
identically satisfied.

The end boundary conditions can be shown as satisfied. T1 = T2 = 0. Considering


the torque on the end, M3 = Mt = 0 since t31 = t32 = 0.

08/07/2017 Dr. Luis Mosquera L. 302


The axial force on the ends is given by

𝑡3𝑖 𝑑𝑆 = 𝑇𝑖
𝒞1

𝑀𝑥1 𝑀
− 𝑑𝑆 = − 𝑥1 𝑑𝑆 = 0
𝒞1 𝐼 𝐼 𝒞1

Providedthat the axes origin is chosen at the centroid of the cross section C. A similar
calculation for M1 gives

𝑀𝑥1 𝑀
𝑀1 = − 𝑥2 𝑑𝑆 = − 𝑥1 𝑥2 𝑑𝑆 = 0
𝒞1 𝐼 𝐼 𝒞1

Finally, 𝑀𝑥1 𝑀𝑥1


𝑀2 = − 𝑥1 𝑑𝑆 = 𝑥12 𝑑𝑆 = 𝑀
𝒞1 𝐼 𝐼 𝒞1

where 𝐼= 𝑥12 𝑑𝑆
𝒞1
08/07/2017 Dr. Luis Mosquera L. 303
Strain components are found from

1
𝜀𝑖𝑗 = 1 + 𝜈 𝑡𝑖𝑗 − 𝜈𝛿𝑖𝑗 𝑡𝑘𝑘
𝐸

𝜈𝑀
𝜀11 = 𝜀22 = 𝑥1
𝐸𝐼
𝑀
𝜀33 = − 𝑥1
𝐸𝐼
𝜀12 = 𝜀13 = 𝜀23 = 0

08/07/2017 Dr. Luis Mosquera L. 304


Flexure

Consider the beam to have end C1 fixed and a flexure load of T acting on C2. The
relaxed boundary conditions are

𝑇1 = 𝑇, 𝑇2 = 𝑇3 = 0, 𝑀𝑖 = 0

Assume the stresses to be


𝑇
𝑡11 = 𝑡22 = 𝑡12 = 0; 𝑡33 = − 𝐿 − 𝑥3 𝑥1
𝐼

With no assumption made on the components t31 and t32. Equilibrium equations are

𝑇
𝑡13,3 = 0, 𝑡23,3 = 0, 𝑡31,1 + 𝑡32,2 = − 𝑥1
𝐼

08/07/2017 Dr. Luis Mosquera L. 305


Which implies

𝑡13 = 𝑡13 𝑥1 , 𝑥2 ; 𝑡23 = 𝑡23 𝑥1 , 𝑥2

Strain components are found from


1
𝜀𝑖𝑗 = 1 + 𝜈 𝑡𝑖𝑗 − 𝜈𝛿𝑖𝑗 𝑡𝑘𝑘
𝐸

𝜈𝑇
𝜀11 = 𝜀22 = 𝐿 − 𝑥3 𝑥1
𝐸𝐼
𝑇
𝜀33 =− 𝐿 − 𝑥3 𝑥1
𝐸𝐼
𝑡23 𝑡31
𝜀23 = ; 𝜀31 = ; 𝜀12 = 0
2𝜇 2𝜇

08/07/2017 Dr. Luis Mosquera L. 306


The compatibility equations results in two nontrivial conditions

𝜕
𝜀 − 𝜀13,2 = 0
𝜕𝑥1 23,1

𝜕 𝜈𝑇
𝜀31,2 − 𝜀23,1 =
𝜕𝑥2 𝐸𝐼

Integration of equations gives

𝜈𝑇
𝜀23,1 − 𝜀31,2 = 𝛼 − 𝑥
𝐸𝐼 2

Where α is taken as a constant

08/07/2017 Dr. Luis Mosquera L. 307


𝜈𝑇
𝜀
The equation, 23,1 − 𝜀31,2 = 𝛼 − 𝑥
𝐸𝐼 2 , may be written as

1 1 1 𝜈𝑇 2
𝜀23 − 𝛼𝑥1 = 𝜀31 + 𝛼𝑥2 − 𝑥2
2 ,1 2 2 𝐸𝐼 ,2

From which the shear strains may be found in terms of a potencial function f

1 1
𝜀32 − 𝛼𝑥1 = 𝑓,2
2 2

1 1 𝜈𝑇 2 1
𝜀31 + 𝛼𝑥2 − 𝑥 = 𝑓
2 2 𝐸𝐼 2 2 ,1

The shear stresses may be written using Hooke’s Law and this equations

08/07/2017 Dr. Luis Mosquera L. 308


𝜈𝜇𝑇 2
𝑡31 = 2𝜇𝜀31 = −𝜇𝛼𝑥2 + 𝑥 + 𝜇𝑓,1
𝐸𝐼 2
𝑡32 = 2𝜇𝜀32 = 𝜇𝛼𝑥1 + 𝜇𝑓,2

The remaining equilibrium condition, is

2
𝑇𝑥1 2 1+𝜈 𝑇
∇ 𝑓 𝑥1 , 𝑥2 =− =− 𝑥1 𝑜𝑛 𝒞
𝜇𝐼 𝐸𝐼

Subjet to the lateral boundary condition 𝑡𝑖𝛼 𝑛𝛼 = 0 in the form

𝜕𝑓 𝜈𝑇 2
= 𝛼 𝑥2 𝑛2 − 𝑥1 𝑛2 − 𝑥 𝑛 𝑜𝑛 𝜕𝒞
𝜕𝑛 𝐸𝐼 2 1

08/07/2017 Dr. Luis Mosquera L. 309


To solve the boundary value problem, define a function F(x1,x2) by

𝑇 𝜈𝑥13 𝜈
𝑓 = 𝛼𝜓 − 𝐹+ + 1 + 𝑥1 𝑥22
𝐸𝐼 6 2

Where ψ is the warping function from torsion

∇2 𝜓 = 0 𝑜𝑛 𝒞;
𝜕𝜓
= 𝑥2 𝑛1 − 𝑥1 𝑛2 𝑜𝑛 𝜕𝒞
𝜕𝑛

Stress components may be written in terms of ψ, α and F as

𝑇 𝜈𝑥12 𝜈
𝑡31 = 𝜇𝛼(𝜓,1 − 𝑥2 ) − 𝐹,1 + + 1 + 𝑥22
2 1+𝜈 𝐼 2 2

08/07/2017 Dr. Luis Mosquera L. 310


𝑇
𝑡32 = 𝜇𝛼(𝜓,2 + 𝑥1 ) − 𝐹 + 2 + 𝜈 𝑥1 𝑥2
2 1 + 𝜈 𝐼 ,2

𝑇
𝑡33 = − (𝐿 − 𝑥3 )𝑥1
𝐼
Subject to the lateral boundary condition

𝜕𝐹 𝜈𝑥12 𝜈
=− + 1 + 𝑥22 𝑛1 − 2 + 𝜈 𝑥1 𝑥2 𝑛2 𝑜𝑛 𝜕𝒞
𝜕𝑛 2 2

08/07/2017 Dr. Luis Mosquera L. 311


P25.- In general, the strain energy density W may be expressed in the form


𝑊 = 𝒞𝛼𝛽 𝜀𝛼 𝜀𝛽 (𝛼, 𝛽 = 1, … . ,6)

Where 𝒞𝛼𝛽∗ is not necessarily symmetric. Show that this equation may be
rearranged to appear in the form

1
𝑊= 𝒞𝛼𝛽 𝜀𝛼 𝜀𝛽
2

Where 𝒞𝛼𝛽 is symmetric, so that now

𝜕𝑊
= 𝒞𝛼𝛽 𝜀𝛽 = 𝑡𝛽
𝜕𝜀𝛽

08/07/2017 Dr. Luis Mosquera L. 312


Plane Elasticity

In a number of engineering applications, specific body geometry and loading


patterns lead to a reduced, essentially two-dimensional form of the equations
of elasticity, and the study of these situations is referred to as plane elasticity.

In plane stress problems, the geometry of the body is that of a thin plate with
one dimension very much smaller than the other two.

The loading in this case is in the plane of the plate and is assumed to be
uniform across the thickness, as shown in figure:

08/07/2017 Dr. Luis Mosquera L. 313


Plane stress

plane strain

FIG (a) Plane stress problems generally involve bodies that are thin in dimensions with
loads perpendicular to that dimension; (b) plane strain problems involve bodies that are
long in one dimension with loads applied along that dimension.
08/07/2017 Dr. Luis Mosquera L. 314
For the plane stress situation, the stress components t33, t31, and t32 are taken as
zero everywhere and the remaining components considered functions of only x1 and
x2. Thus,

𝑡𝑖𝑗 = 𝑡𝑖𝑗 𝑥1 , 𝑥2 (𝑖, 𝑗 = 1,2)

and as a result, the equilibrium equations, 𝑡𝑗𝑖 ,𝑗 + 𝜌𝑏𝑖 = 0, reduce to the specific
equations
𝑡11,1 + 𝑡12,2 + 𝜌𝑏1 = 0

𝑡21,1 + 𝑡22,2 + 𝜌𝑏2 = 0

The strain-displacement relations, become


𝜀11 = 𝑢1,1

𝜀22 = 𝑢2,2

2𝜀12 = 𝑢1,2 + 𝑢2,1


08/07/2017 Dr. Luis Mosquera L. 315
and at the same time the strain compatibility equations,

𝜀11,22 + 𝜀22,11 = 2𝜀12,12

Hooke’s law equations for plane stress are written

1
𝜀11 = 𝑡 − 𝜈𝑡22
𝐸 11
1
𝜀22 = 𝑡 − 𝜈𝑡11
𝐸 22
1+𝜈 𝑡12
𝜀12 = 𝑡12 =
𝐸 2𝐺

along with

𝜈 −𝜈
𝜀33 = − 𝑡11 + 𝑡22 = 𝜀11 + 𝜀22
𝐸 1−𝜈

08/07/2017 Dr. Luis Mosquera L. 316


By inverting the equations, we express the stress components in terms of the
strains as
𝐸
𝑡11 = 2
𝜀11 + 𝜈𝜀22
1−𝜈
𝐸
𝑡22 = 𝜀 + 𝜈𝜀11
1 − 𝜈 2 22
𝐸
𝑡12 = 𝜀
1 + 𝜈 12
These equations may be conveniently cast into the matrix formulation

𝑡11 1 𝜈 0 𝜀11
𝐸
𝑡22 = 𝜀22
1 − 𝜈 2 𝜈 1 0
𝑡12 0 0 1−𝜈 𝜀12

08/07/2017 Dr. Luis Mosquera L. 317


In terms of the displacement components, ui (i = 1, 2), the plane stress field
equations may be combined to develop a Navier-type equation for elastostatics,
namely,

𝐸 𝐸
𝑢 + 𝑢 + 𝜌𝑏𝑖 = 0 ; (𝑖, 𝑗 = 1,2)
2 1 + 𝜈 𝑖,𝑗𝑗 2 1 − 𝜈 𝑗 ,𝑗𝑖

08/07/2017 Dr. Luis Mosquera L. 318


For the plane strain situation, figure b, we assume that u3 = 0 and that the remaining
two displacement components are functions of only x1 and x2,

𝑢𝑖 = 𝑢𝑖 𝑥1 , 𝑥2 ; (𝑖 = 1,2)

Here, Hooke’s law may be written in terms of engineering constants as

𝐸
𝑡11 = 1 − 𝜈 𝜀11 + 𝜈𝜀22
1 + 𝜈 1 − 2𝜈
𝐸
𝑡22 = 1 − 𝜈 𝜀22 + 𝜈𝜀11
1 + 𝜈 1 − 2𝜈
𝐸
𝑡12 = 𝜀
1 + 𝜈 12

08/07/2017 Dr. Luis Mosquera L. 319


along with
𝐸𝜈
𝑡33 = 𝜀11 + 𝜀22 = 𝜈 𝑡11 − 𝑡22
1 + 𝜈 1 − 2𝜈

The first three of these equations may be expressed in matrix form by

𝑡11 1−𝜈 𝜈 0 𝜀11


𝐸
𝑡22 = 𝜈 1−𝜈 0 𝜀22
𝑡12 1 + 𝜈 1 − 2𝜈 𝜀12
0 0 1 − 2𝜈

Furthermore, by inverting the same three equations, we may express Hooke’s


law for plane strain by the equations

1+𝜈
𝜀11 = (1 − 𝜈)𝑡11 − 𝜈𝑡22
𝐸
1+𝜈
𝜀22 = 1 − 𝜈 𝑡22 + 𝜈𝑡11
𝐸
1+𝜈 2(1 + 𝜈) 𝑡12 𝑡12
𝜀12 = 𝑡 = =
𝐸 12 𝐸 2 2𝐺
08/07/2017 Dr. Luis Mosquera L. 320
By combining the field equations with Hooke’s law for elastostatic plane strain, we
obtain the appropriate Navier equation as

𝐸 𝐸
𝑢 + 𝑢 + 𝜌𝑏𝑖 = 0 ; (𝑖, 𝑗 = 1,2)
2 1 + 𝜈 𝑖,𝑗𝑗 2 1 + 𝜈 (1 − 2𝜈) 𝑗 ,𝑗𝑖

08/07/2017 Dr. Luis Mosquera L. 321


Airy Stress Function

When body forces are conservative with a potential function V = V(x1, x2) such that
bi = –V,ii, we may introduce the Airy stress function, φ = φ(x1, x2) in terms of which
the stresses are given by
𝑡11 = 𝜙,22 + 𝜌𝑉

𝑡22 = 𝜙,11 + 𝜌𝑉

𝑡12 = −𝜙,12

Note that by using this definition the equilibrium equations are satisfied
identically.

For the case of plane stress we obtain

𝑡11,22 + 𝑡22,11 − 𝜈 𝑡11,11 + 𝑡22,22 = 2(1 + 𝜈)𝑡12,12

08/07/2017 Dr. Luis Mosquera L. 322


which in terms of φ becomes

𝜙,1111 + 2𝜙,1212 + 𝜙,2222 = − 1 − 𝜈 𝜌 𝑉,11 + 𝑉,22

Similarly, for the case of plane strain, the result is

1 − 𝜈 𝑡11,22 + 𝑡22,11 − 𝜈 𝑡11,11 + 𝑡22,22 = 2𝑡12,12

or in terms of φ

𝜙,1111 + 2𝜙,1212 + 𝜙,2222 = − 1 − 2𝜈 𝜌 𝑉,11 − 𝑉,22 /(1 − 𝜈)

If the body forces consist of gravitational forces only, or if they are constant forces,
the right-hand sides of both equations reduce to zero and φ must then satisfy the bi-
harmonic equation

𝜙,1111 + 2𝜙,1212 + 𝜙,2222 = ∇4 𝜙 = 0

08/07/2017 Dr. Luis Mosquera L. 323


Example

For a thin rectangular plate of the dimensions shown in Figure a, consider the general
polynomial of the third degree as the Airy stress function and from it determine the
stresses. Assume body forces are zero.

08/07/2017 Dr. Luis Mosquera L. 324


Select a polynomial stress function of the form 𝜙3 = 𝐴3 𝑥13 + 𝐵3 𝑥12 𝑥2 + 𝐶3 𝑥1 𝑥22 + 𝐷3 𝑥23
Choosing this particular polynomial form for the stress function is not arbitrary; the
choice is based on many trials of different order polynomials.

By direct substitution into equation

𝜙,1111 + 2𝜙,1212 + 𝜙,2222 = ∇4 𝜙 = 0

we confirm that φ3 is bi-harmonic. Further, the stresses are given as

𝑡11 = 2𝐶3 𝑥1 + 6𝐷3 𝑥2

𝑡22 = 6𝐴3 𝑥1 + 2𝐵3 𝑥2

𝑡12 = −2𝐵3 𝑥1 − 2𝐶3 𝑥2

By selecting different constants to be zero and nonzero, different physical problems


may be solved. Here, two specific cases will be considered.

08/07/2017 Dr. Luis Mosquera L. 325


(a) Assume all coefficients in φ3 are zero except D3. This may be shown to solve the case
of pure bending of a beam by equilibrating moments on the ends. Stress in the fiber
direction of the beam varies linearly with the distance from the x1 axis

𝑡11 = 6𝐷3 𝑥2 ; 𝑡22 = 𝑡12 = 0

as is the case for simple bending. Similarly, by taking only A3 as nonzero, the solution
is for bending moments applied to a beam whose lengthwise direction is taken to be
x2 rather than x1 direction.

(b) If only B3 (or C3) is non-zero, both shear and normal stresses are present. Figure (c)
shows the stress pattern for B3 ≠ 0.

08/07/2017 Dr. Luis Mosquera L. 326


Example

Consider a special stress function having the form

𝜙 ∗ = 𝐵2 𝑥1 𝑥2 + 𝐷4 𝑥1 𝑥23

Show that this stress function may be adapted to solve for the stresses in an end-
loaded cantilever beam shown in the sketch. Assume the body forces are zero for
this problem.

08/07/2017 Dr. Luis Mosquera L. 327


Solution

It is verified, by direct substitution, that ∇4 ∅∗ = 0 .The stress components


are directly computed as
𝑡11 = 6𝐷4 𝑥1 𝑥2

𝑡22 = 0

𝑡12 = −𝐵2 − 3𝐷4 𝑥22

In order for the top and bottom surfaces of the beam to be stress free, t12 must be
zero at x2 = ± c. Using this condition B2 is determined in terms of D4 as B2 = –3D4 c2.
The shear stress is thus given in terms of single constant B2

𝐵2 𝑥22
𝑡12 = −𝐵2 + 2
𝑐

The concentrated load is modeled as the totality of the shear stress t12 on the free
end of the beam. Thus, the result of integrating this stress over the free end of the
beam at x1 = 0 yields the applied force P. In equation form
08/07/2017 Dr. Luis Mosquera L. 328
𝑐
𝑥22
𝑃=− −𝐵2 + 𝐵2 2 𝑑𝑥2
−𝑐 𝑐

where the minus sign is required due to the sign convention on shear stress. Carrying
out the integration we have B2 = 3P/4c so that stress components may now be written
as
3𝑃
𝑡11 = − 3 𝑥1 𝑥2
2𝑐
𝑡22 = 0

3𝑃 𝑥22
𝑡12 =− 1− 2
4𝑐 𝑐

But for this beam the plane moment of inertia of the cross section is I = 2c3/3
so that now
𝑃
𝑡11 = − 𝑥1 𝑥2
𝐼
𝑡22 = 0
𝑃 2
𝑡12 = − 𝑐 − 𝑥22
08/07/2017 2𝐼
Dr. Luis Mosquera L. 329
The Airy stress function expressed in terms of polar coordinates

We introduce here the basic material element together with the relevant stress
components in terms of the coordinates r and θ as shown on figure.

08/07/2017 Dr. Luis Mosquera L. 330


Using this element and summing forces in the radial direction results in the equilibrium
equation

𝜕𝑡𝑟𝑟 1 𝜕𝑡𝑟𝜃 𝑡𝑟𝑟 − 𝑡𝜃𝜃


+ + +𝑅 =0
𝜕𝑟 𝑟 𝜕𝜃 𝑟

and summing forces tangentially yields

1 𝜕𝑡𝜃𝜃 𝜕𝑡𝑟𝜃 𝑡𝑟𝜃


+ +2 +Θ=0
𝑟 𝜕𝜃 𝜕𝑟 𝑟

in which R and Θ represent body forces. In the absence of such forces this equations
are satisfied by

1 𝜕𝜙 1 𝜕 2 𝜙
𝑡𝑟𝑟 = +
𝑟 𝜕𝑟 𝑟 2 𝜕𝜃 2
𝜕2 𝜙
𝑡𝜃𝜃 = 2
𝜕𝑟
1 𝜕𝜙 1 𝜕 2 𝜙 𝜕 1 𝜕𝜙
𝑡𝑟𝜃 = 2 − =−
𝑟 𝜕𝜃 𝑟 𝜕𝑟𝜕𝜃 𝜕𝑟 𝑟 𝜕𝜃
08/07/2017 Dr. Luis Mosquera L. 331
To qualify as an Airy stress function φ must once again satisfy the condition ∇4 ϕ = 0
which in polar form is

4
𝜕2 1 𝜕 1 𝜕2 𝜕 2 𝜙 1 𝜕𝜙 1 𝜕 2 𝜙
∇ ϕ= + + + + =0
𝜕𝑟 2 𝑟 𝜕𝑟 𝑟 2 𝜕𝜃 2 𝜕𝑟 2 𝑟 𝜕𝑟 𝑟 2 𝜕𝜃 2

For stress fields symmetrical to the polar axis this equation reduces to

4
𝜕2 1 𝜕 𝜕 2 𝜙 1 𝜕𝜙
∇ ϕ= + + =0
𝜕𝑟 2 𝑟 𝜕𝑟 𝜕𝑟 2 𝑟 𝜕𝑟

The general solution to this differential equation is given by

𝜙 = 𝐴 𝑙𝑛𝑟 + 𝐵𝑟 2 𝑙𝑛𝑟 + 𝐶𝑟 2 + 𝐷

so that for the symmetrical case the stress components take the form

08/07/2017 Dr. Luis Mosquera L. 332


1 𝜕𝜙 𝐴
𝑡𝑟𝑟 = = + 𝐵 1 + 2 𝑙𝑛𝑟 + 2𝐶
𝑟 𝜕𝑟 𝑟 2
𝜕2 𝜙 𝐴
𝑡𝜃𝜃 = 2 = − 2 + 𝐵 3 + 2 𝑙𝑛𝑟 + 2𝐶
𝜕𝑟 𝑟
𝑡𝑟𝜃 = 0

When there is no hole at the origin in the elastic body under consideration, A and B
must be zero since otherwise infinite stresses would result at that point. Thus, for a
plate without a hole only uniform tension or compression can exist as a symmetrical
case.

08/07/2017 Dr. Luis Mosquera L. 333


Example

Determine the stresses in a curved beam of the dimensions shown when subjected
to constant equilibrating moments.

08/07/2017 Dr. Luis Mosquera L. 334


Solution

Boundary conditions require

𝑡𝑟𝑟 = 0 𝑎𝑡 𝑟 = 𝑎, 𝑎𝑛𝑑 𝑎𝑡 𝑟 = 𝑏,
𝑏
𝑡𝜃𝜃 𝑑𝑟 = 0 𝑜𝑛 𝑡ℎ𝑒 𝑒𝑛𝑑 𝑓𝑎𝑐𝑒𝑠
𝑎

𝑏
𝑟𝑡𝑟𝑟 𝑑𝑟 = −𝑀 𝑜𝑛 𝑡ℎ𝑒 𝑒𝑛𝑑 𝑓𝑎𝑐𝑒𝑠
𝑎

𝑡𝑟𝜃 = 0 𝑒𝑣𝑒𝑟𝑦𝑤ℎ𝑒𝑟𝑒 𝑜𝑛 𝑡ℎ𝑒 𝑏𝑜𝑢𝑛𝑑𝑎𝑟𝑦

The inner and outer radii are free of normal stress which can be written in terms
of boundary condition as
𝐴
+ 𝐵 1 + 2 𝑙𝑛𝑎 + 2𝐶 = 0,
𝑎2
𝐴
+ 𝐵 1 + 2 𝑙𝑛𝑏 + 2𝐶 = 0
𝑏2
08/07/2017 Dr. Luis Mosquera L. 335
No transverse loading is present on the ends of the curved beam which may
be written in terms of boundary condition as

𝑏 𝑏 𝑏
𝜕2 𝜙 𝜕𝜙
𝑡𝜃𝜃 𝑑𝑟 = 𝑑𝑟 = =0
𝑎 𝑎 𝜕𝑟 2 𝜕𝑟 𝑎

Finally, the applied moments on the ends may be written in terms of boundary
condition
𝑏 𝑏 𝑏
𝜕2 𝜙 𝜕𝜙 𝜕𝜙
𝑟 2 𝑑𝑟 = (𝑟 ) − 𝑑𝑟 = −𝑀
𝑎 𝜕𝑟 𝜕𝑟 𝑎 𝑎 𝜕𝑟

Because of condition (1) the bracketed term here is zero and from the integral
term
𝑏
𝐴 𝑙𝑛 + 𝐵 𝑏 2 𝑙𝑛𝑏 − 𝑎2 𝑙𝑛𝑎 + 𝐶 𝑏 2 − 𝑎2 = 𝑀
𝑎

08/07/2017 Dr. Luis Mosquera L. 336


Therefore,
4𝑀 2 2 𝑏
𝐴=− 𝑎 𝑏 𝑙𝑛
𝑁 𝑎
2𝑀 2
𝐵=− 𝑏 − 𝑎2
𝑁
𝑀 2
𝐶= 𝑏 − 𝑎2 + 2 𝑏 2 𝑙𝑛𝑏 − 𝑎2 𝑙𝑛𝑎
𝑁
where
𝑁 = 𝑏 2 − 𝑎2 2
− 4𝑎2 𝑏2 𝑙𝑛 𝑏/𝑎 2

Finally, the stress components may be written

4𝑀 𝑎2 𝑏 2 2
𝑟 2
𝑎
𝑡𝑟𝑟 =− 𝑙𝑛 𝑏/𝑎 + 𝑏 𝑙𝑛 + 𝑎 𝑙𝑛
𝑁 𝑟2 𝑏 𝑟

4𝑀 𝑎2 𝑏 2 𝑏 𝑟 𝑎
𝑡𝜃𝜃 =− − 2 𝑙𝑛 + 𝑏 2 𝑙𝑛 + 𝑎2 𝑙𝑛 + 𝑏 2 − 𝑎2
𝑁 𝑟 𝑎 𝑏 𝑟

𝑡𝑟𝜃 = 0
08/07/2017 Dr. Luis Mosquera L. 337
Three-Dimensional Elasticity

Solutions of three-dimensional elasticity problems focus on two distinct formulations.


First, the displacement formulation is based upon solutions of the Navier equations.
The second formulation, called the stress formulation, utilizes solutions of the
equilibrium equations in association with the Beltrami-Michell stress equations and
based upon the compatibility equations in terms of strains.

Starting with the fundamental equations of elastostatics, we have:

𝑡𝑗𝑖 ,𝑗 + 𝜌𝑏𝑖 = 0

2𝜀𝑖𝑗 = 𝑢𝑖,𝑗 + 𝑢𝑗 ,𝑖
𝑡𝑖𝑗 = 𝜆𝛿𝑖𝑗 𝜀𝑘𝑘 + 2𝜇𝜀𝑖𝑗
1
𝜀𝑖𝑗 = 1 + 𝜈 𝑡𝑖𝑗 − 𝜈𝛿𝑖𝑗 𝑡𝑘𝑘
𝐸

08/07/2017 Dr. Luis Mosquera L. 338


The above equations,

𝜇𝑢𝑖,𝑗𝑗 + 𝜆 + 𝜇 𝑢𝑗 ,𝑗𝑖 + 𝜌𝑏𝑖 = 0

which comprise three second-order partial differential equations known as the


Navier equations. For the stress formulation we convert the strain equations
of compatibility and repeated here as

𝜀𝑖𝑗 ,𝑘𝑚 + 𝜀𝑘𝑚 ,𝑖𝑗 − 𝜀𝑖𝑘,𝑗𝑚 − 𝜀𝑗𝑚 ,𝑖𝑘 = 0

into the equivalent expression in terms of stresses, and combine that result with
above equaions to obtain
1 𝜈
𝑡𝑖𝑗 ,𝑘𝑘 + 𝑡𝑘𝑘 ,𝑖𝑗 + 𝜌 𝑏𝑖,𝑗 + 𝑏𝑗 ,𝑖 + 𝛿𝑖𝑗 𝜌𝑏𝑘,𝑘 = 0
1+𝜈 1+𝜈

which are the Beltrami-Michell equations of compatibility.

08/07/2017 Dr. Luis Mosquera L. 339


Let us first consider solutions developed through the displacement formulation. By
inserting an expression for ui in terms of the proposed potentials into the Navier
equations we obtain the governing equations for the appropriate potentials.

The method used in our first approach rests upon the well-known theorem of
Helmholtz which states that any vector function that is continuous and finite, and
which vanishes at infinity, may be resolved into a pair of components: one a rotation
vector, the other an irrotational vector. Thus, if the curl of an arbitrary vector a is zero,
then a is the gradient of a scalar φ, and a is irrotational, or as it is sometimes called,
solenoidal. At the same time, if the divergence of the vector a is zero, then a is the
curl of another vector ψ, and is a rotational vector.

Accordingly, in keeping with the Helmholtz theorem, we assume that the


displacement field is given by

𝑢𝑖 = 𝜙,𝑖 + 𝜀𝑖𝑝𝑞 𝜓𝑞,𝑝

08/07/2017 Dr. Luis Mosquera L. 340


where φ,i is representative of the irrotational portion, and curl ψ the rotational
portion. Substituting this displacement vector with bi taken as zero, namely

𝜇𝑢𝑖,𝑗𝑗 + 𝜆 + 𝜇 𝑢𝑗 ,𝑗𝑖 = 0

we obtain
𝜇𝜙,𝑖𝑗𝑗 + 𝜇𝜀𝑖𝑝𝑞 𝜓𝑞,𝑝𝑗𝑗 + 𝜆 + 𝜇 𝜙,𝑖𝑗𝑗 + 𝜆 + 𝜇 𝜀𝑗𝑝𝑞 𝜓𝑞,𝑝𝑗𝑖 = 0

which reduces to
𝜆 + 2𝜇 𝜙,𝑖𝑗𝑗 + 𝜇𝜀𝑖𝑝𝑞 𝜓𝑞,𝑝𝑗𝑗 = 0

since 𝜀𝑗 𝑝𝑞 𝜓𝑞,𝑝𝑗𝑖 = 0

In coordinate-free notation

𝜆 + 2𝜇 ∇∇2 𝜙 + 𝜇∇ × ∇2 𝜓 = 0

08/07/2017 Dr. Luis Mosquera L. 341


Any set of φ and ψ which satisfies the above equations provides a displacement
field satisfying the Navier equation. Clearly, one such set is obtained by requiring φ
and ψ to be harmonic:

∇2 𝜙 = 0

∇2 𝜓 = 0

Note that, if we choose ∇2 𝜙 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 and ψ = 0, the scalar function φ is known as


the Lamé strain potential. By taking the divergence of this equation, and remembering
that the divergence of a curl vanishes, we see that

∇4 𝜙 = 0

is a solution of the resulting equation so that a bi-harmonic function as φ also yields a


solution for ui. Similarly, we find that

∇4 𝜓 = 0

also provides for a solution ui.

08/07/2017 Dr. Luis Mosquera L. 342


The second approach for solving the Navier equations is based on the premise of
expressing the displacement field in terms of the second derivatives of a vector
known as the Galerkin vector, and designated here by 𝐹 = 𝐹𝑖 𝑒𝑖

In this approach we assume the displacement ui is given in terms of the Galerkin


vector specifically by the equation

𝑢𝑖 = 2 1 − 𝜈 𝐹𝑖,𝑗𝑗 − 𝐹𝑗 ,𝑗𝑖

which is substituted directly into Eq 𝜇𝑢𝑖,𝑗𝑗 + 𝜆 + 𝜇 𝑢𝑗 ,𝑗𝑖 = 0

and reducing the resulting equations with the help of the identity 𝜆 = 2𝜈𝜇/ 1 − 2𝜈

we find that the Navier equations are satisfied if

∇4 𝐹 = 0
Thus, any bi-harmonic vector is suitable as a Galerkin vector.
08/07/2017 Dr. Luis Mosquera L. 343
there is a relationship between φ and ψ with F. It can be been shown that

𝜙 = −𝐹𝑖,𝑖

and
𝜀𝑖𝑗𝑘 𝜓𝑘,𝑗 = 2 1 − 𝜈 𝐹𝑖,𝑗𝑗

If Fi is not only bi-harmonic, but harmonic as well, this equation reduces to

𝜀𝑖𝑗𝑘 𝜓𝑘,𝑗 = 0

and the relationship between φ and Fi becomes

𝜙,𝑖𝑖 = −𝐹𝑖,𝑗𝑗 𝑖

In this case φ is called the Lamé strain potential.

08/07/2017 Dr. Luis Mosquera L. 344


Example

Consider a Galerkin vector of the form 𝐹 = 𝐹𝑖 𝑒𝑖 where F3 is a function of the


coordinates, that is 𝐹3 = 𝐹3 (𝑥1 , 𝑥2 , 𝑥3 ) . Apply this vector to obtain the solution
to the problem of a concentrated force acting at the origin of coordinates in the
direction of the positive x3 axis of a very large elastic body. This is called the Kelvin
problem.

Solution

Let 𝑢𝑖 = 2 1 − 𝜈 𝐹𝑖,𝑗𝑗 − 𝐹𝑗 ,𝑗𝑖

Therefore,
𝑢1 = −𝐹3,31

𝑢2 = −𝐹3,32

𝑢3 = 2 1 − 𝜈 𝐹3,11 + 𝐹3,22 + 𝐹3,33 − 𝐹3,33


08/07/2017 Dr. Luis Mosquera L. 345
Take F3 to be proportional to the distance squared from the origin as defined by
F3 = BR where B is a constant and 𝑅 2 = 𝑥12 + 𝑥22 + 𝑥32 thus, the displacements

are

𝐵𝑥3 𝑥1
𝑢1 =
𝑅3
𝐵𝑥3 𝑥2
𝑢2 =
𝑅3
4 1−𝜈 𝑥12 + 𝑥22
𝑢3 = 𝐵 −
𝑅 𝑅3

From these displacement components the stresses may be computed using Hooke’s
law.

08/07/2017 Dr. Luis Mosquera L. 346


Prob 26.- For an isotropic elastic medium as defined by 𝑡𝑖𝑗 = 𝜆𝛿𝑖𝑗 𝜀𝑘𝑘 + 2𝜇𝜀𝑖𝑗 ,
express the strain energy density in terms of
(a) the components of εij
(b) the components of σij
(c) the invariants of εij

Prob 27.- Show for an isotropic elastic medium that

1 2 𝜆+𝜇
𝑎) =
1 + 𝜈 3𝜆 + 2𝜇
𝜈 𝜆
𝑏) =
1 + 𝜈 𝜆 + 2𝜇

2𝜇𝜈 3𝐾𝜈
𝑐) =
1 − 2𝜈 1 + 𝜈

𝑑) 2𝜇 1 + 𝜈 = 3𝐾 1 − 2𝜈

08/07/2017 Dr. Luis Mosquera L. 347


Prob 28.- Develop a formula in terms of the strain components for the strain energy
density W for the case of an orthotropic elastic medium.

Prob 29.- A rectangular beam of width unity and length 2L carries a uniformly
distributed load of q lb/ft as shown. Shear forces V support the beam at both ends.
List the six boundary conditions for this beam the stresses must satisfy.

08/07/2017 Dr. Luis Mosquera L. 348


Classical Fluids

Viscous Stress Tensor, Stokesian, and Newtonian Fluids

A fundamental characteristic of any fluid is that the action of shear stresses, no


matter how small they may be, will cause the fluid to deform continuously as long
as the stresses act.
This implies that the stress vector on an arbitrary element of surface at any point
in a fluid at rest is proportional to the normal ni of that element, but independent
of its direction:
𝑛
𝑡𝑖 = 𝑡𝑖𝑗 𝑛𝑗 = −𝑝0 𝑛𝑖

where the (positive) proportionality constant po is the thermostatic pressure or, as it


is frequently called, the hydrostatic pressure. We note that

𝑡𝑖𝑗 = −𝑝0 𝛿𝑖𝑗

which indicates that every direction is a principal stress direction at any point, and
that the hydrostatic pressure is equal to the mean normal stress,

08/07/2017 Dr. Luis Mosquera L. 349


1
𝑝0 = − 𝑡𝑖𝑖
3
For a fluid in motion the shear stresses are not usually zero, and in this case

𝑡𝑖𝑗 = −𝑝𝛿𝑖𝑗 + 𝜏𝑖𝑗

Where τij is called the viscous stress tensor

For a fluid in motion, p is not equal to the mean normal stress, but instead is
given by
1
𝑝 = − 𝑡𝑖𝑖 − 𝜏𝑖𝑖
3

We assume that τij is a function of the rate of deformation tensor Dij

𝜏𝑖𝑗 = 𝑓𝑖𝑗 (𝐷)

08/07/2017 Dr. Luis Mosquera L. 350


If the functional relationship in this equation is nonlinear, the fluid is called a
Stokesian fluid.
When fij defines τij as a linear function of Dij, the fluid is known as a Newtonian
fluid, and we represent it by the equation:

𝜏𝑖𝑗 = 𝐾𝑖𝑗 𝑝𝑞 𝑑𝑝𝑞

for a homogeneous, isotropic Newtonian fluid, the constitutive equation is

𝑡𝑖𝑗 = −𝑝𝛿𝑖𝑗 + 𝜆∗ 𝛿𝑖𝑗 𝑑𝑘𝑘 + 2𝜇∗ 𝑑𝑖𝑗

where λ* and μ* are viscosity coefficients

From this equation we see that the mean normal stress for a Newtonian fluid is

1 1
𝑡𝑖𝑖 = −𝑝 + 3𝜆∗ + 2𝜇∗ 𝑑𝑖𝑗 = −𝑝 + 𝑘 ∗ 𝑑𝑖𝑖
3 3

08/07/2017 Dr. Luis Mosquera L. 351


Where k* is known as the coefficient of bulk viscosity. The condition

1
𝑘 = 3𝜆∗ + 2𝜇∗

3
is known as Stokes condition. If we introduce the deviator tensors
.
1
𝑆𝑖𝑗 = 𝑡𝑖𝑗 − 𝛿𝑖𝑗 𝑡𝑘𝑘
3
1
𝛽𝑖𝑗 = 𝑑𝑖𝑗 − 𝛿𝑖𝑗 𝑑𝑘𝑘
3
for stress and for rate of deformation, we obtain

1 1
𝑆𝑖𝑗 + 𝛿𝑖𝑗 𝑡𝑘𝑘 = −𝑝𝛿𝑖𝑗 + 3𝜆∗ + 2𝜇∗ 𝛿𝑖𝑗 𝑑𝑘𝑘 + 2𝜇∗ 𝛽𝑖𝑗
3 3

08/07/2017 Dr. Luis Mosquera L. 352


which may be conveniently split into the pair of constitutive equations

𝑆𝑖𝑗 = 2𝜇∗ 𝛽𝑖𝑗

𝑡𝑖𝑖 = −3(𝑝 − 𝑘 ∗ 𝑑𝑘𝑘 )

Basic Equations of Viscous Flow, Navier-Stokes Equations

because the viscous forces are related directly to the velocity field, it is
customary to employ the Eulerian description in writing the governing equations
for boundary value problems in viscous fluid theory. Thus, for the
thermomechanical behavior of a Newtonian fluid, the following field equations
must be satisfied:

(a) the continuity equation

𝜌 + 𝜌𝑣𝑖,𝑖 = 0
08/07/2017 Dr. Luis Mosquera L. 353
(b) the equations of motion

𝑡𝑖𝑗 ,𝑗 + 𝜌𝑏𝑖 = 𝜌𝑣𝑖

(c) the constitutive equations

𝑡𝑖𝑗 = −𝑝𝛿𝑖𝑗 + 𝜆∗ 𝛿𝑖𝑗 𝑑𝑘𝑘 + 2𝜇∗ 𝑑𝑖𝑗

(d) the energy equation

𝜌𝑢 = 𝑡𝑖𝑗 𝑑𝑖𝑗 − 𝑞𝑖,𝑖 + 𝜌𝑟

(e) the kinetic equation of state

𝑝 = 𝑝(𝜌, 𝜃)

08/07/2017 Dr. Luis Mosquera L. 354


(f) the caloric equation of state

𝑢 = 𝑢(𝜌, 𝜃)

(g) the heat conduction equation

𝑞𝑖 = −𝑘𝜃,𝑖

This system of equations together with the definition of the rate of deformation
tensor,
1
𝑑𝑖𝑗 = 𝑣𝑖,𝑗 + 𝑣𝑗 ,𝑖
2

represents 22 equations in the 22 unknowns, tij, ρ, νi, dij, u, qi, p and θ.

08/07/2017 Dr. Luis Mosquera L. 355


Certain of the above field equations may be combined to offer a more compact
formulation of viscous fluid problems. Thus, we can obtain


𝜌𝑣𝑖 = 𝜌𝑏𝑖 − 𝑝,𝑖 + 𝜆 + 𝜇∗ 𝑣𝑗,𝑗𝑖 + 𝜇∗ 𝑣𝑖,𝑗𝑗

which are known as the Navier-Stokes equations for fluids.

In the Eulerian formulation


𝜕𝑣𝑖
𝑣𝑖 = + 𝑣𝑗 𝑣𝑖,𝑗
𝜕𝑡
If Stokes condition 2
𝜆∗ = − 𝜇 ∗ is assumed, the navier equation reduces to
the form 3

1 ∗
𝜌𝑣𝑖 = 𝜌𝑏𝑖 − 𝑝,𝑖 + 𝜇 (𝑣𝑗,𝑗𝑖 + 3𝑣𝑖,𝑗𝑗 )
3

08/07/2017 Dr. Luis Mosquera L. 356


Steady Flow, Irrotational Flow, Potential Flow

If the velocity components of a fluid are independent of time, the motion is called
a steady flow. In such cases, the material derivative of the velocity,

𝜕𝑣𝑖
𝑣𝑖 = + 𝑣𝑗 𝑣𝑖,𝑗
𝜕𝑡
reduces to the simpler form

𝑣𝑖 = 𝑣𝑗 𝑣𝑖,𝑗

Thus, for a steady flow, the Euler equation is modified to read

𝜌𝑣𝑗 𝑣𝑖,𝑗 = 𝜌𝑏𝑖 − 𝑝,𝑖

08/07/2017 Dr. Luis Mosquera L. 357


Furthermore, if the velocity field is constant and equal to zero everywhere, the
fluid is at rest and the theory for this condition is called hydrostatics. For this, the
Navier-Stokes equations are simply

𝜌𝑏𝑖 − 𝑝,𝑖 = 0

Assuming a barotropic condition between ρ and p , it is possible to define a pressure


function in the form
𝑝
𝑑𝑝
𝑃 𝑝 =
𝑝0 𝜌

In addition, if the body forces are conservative, we may express them in terms of a
scalar potential function Ω by the relationship

𝑏𝑖 = −Ω,𝑖

08/07/2017 Dr. Luis Mosquera L. 358


Therefore, it follow that

1 ∇𝑝
𝑃,𝑖 = 𝑝,𝑖 𝑜𝑟 ∇𝑃 =
𝜌 𝜌

so that now may be written

Ω+𝑃 ,𝑖 =0

as the governing equation for steady flow of a barotropic fluid with conservative
body forces.

08/07/2017 Dr. Luis Mosquera L. 359


Example

An incompressible Newtonian fluid maintains a steady flow under the action of


gravity down an inclined plane of slope β. If the thickness of the fluid
perpendicular to the plane is h and the pressure on the free surface is p = po
(a constant), determine the pressure and velocity fields for this flow.

08/07/2017 Dr. Luis Mosquera L. 360


Solution

Assume v1 = v3 = 0. v2 = v2(x2, x3). By the continuity equation for incompresible


flow, vi,i = 0. Hence, v2,2 = 0 and v2 = v2(x3). Thus, the rate of deformation tensor
1 𝜕𝑣2
has components 𝑑23 = 𝑑32 = and all others equal to zero. The
2 𝜕𝑥3
Newtonian constitutive equation is given in this case by

𝑡𝑖𝑗 = −𝑝𝛿𝑖𝑗 + 2𝜇∗ 𝑑𝑖𝑗

from which we calculate

−𝑝 0 0
𝜕𝑣2
0 −𝑝 𝜇∗
𝑡𝑖𝑗 = 𝜕𝑥3
𝜕𝑣2
0 𝜇∗ −𝑝
𝜕𝑥3

08/07/2017 Dr. Luis Mosquera L. 361


the equations of motion having the steady flow form

𝑡𝑖𝑗 ,𝑗 + 𝜌𝑏𝑖 = 𝜌𝑣𝑗 𝑣𝑖,𝑗

result in component equations

(for i = 1) −𝑝,1 = 0
2
∗ 𝜕 𝑣2
(for i = 2) −𝑝,2 + 𝜇 + 𝜌𝑔𝑠𝑖𝑛𝛽 = 0
𝜕𝑥32

(for i = 3) −𝑝,3 − 𝜌𝑔𝑐𝑜𝑠𝛽 = 0

Because gravity is the only body force,

𝒃 = 𝑔 𝑠𝑖𝑛𝛽𝑒2 − 𝑐𝑜𝑠𝛽𝑒3

08/07/2017 Dr. Luis Mosquera L. 362


Integrating the equations gives

𝑝 = − 𝜌𝑔𝑐𝑜𝑠𝛽 𝑥3 + 𝑓(𝑥2 )

At the free surface (x3 = h), p = po, and so

𝑓 𝑥2 = 𝑝0 + 𝜌𝑔ℎ𝑐𝑜𝑠𝛽

and thus
𝑝 = 𝑝0 + 𝜌𝑔𝑐𝑜𝑠𝛽 ℎ − 𝑥3

Next, by integrating the middle equation above (for i = 2) twice with respect to x3,
we obtain

−𝜌𝑔𝑠𝑒𝑛𝛽 2
𝑣2 = 𝑥3 + 𝑎𝑥3 + 𝑏
2𝜇 ∗

08/07/2017 Dr. Luis Mosquera L. 363


with a and b constants of integration. But from the boundary conditions,

1. v2 = 0 when x3 = 0, therefore b = 0

𝜌𝑔ℎ
2. σ23 = 0 when x3 = h, therefore 𝑎 = ∗ 𝑠𝑒𝑛𝛽
𝜇

Finally,
𝜌𝑔𝑠𝑒𝑛𝛽
𝑣2 = 2ℎ − 𝑥3 𝑥3
2𝜇 ∗

08/07/2017 Dr. Luis Mosquera L. 364