Theory of Elasticity

Supplementary Notes
J. N. Reddy
Distinguished Professor
Department of Mechanical Engineering
Texas A&M University, College Station
Texas, USA 77843–3123

@Copyright: J. N. Reddy, 2003

 J. N. REDDY 1

1. Introduction
Consider a deformable body of known geometry, constitution, and loading.
For a given geometry and loading, the body will undergo deformation (i.e.,
macroscopic geometric changes within the body). If the applied loads are
time-dependent, the deformation of the body will be a function of time, i.e.,
the geometry of the body will change continuously with time. If the loads are
applied slowly so that the deformation is only dependent on the loads, the
body will take a definitive shape at the end of each load application. Whether
the deformation is time dependent or not, the forces in the deformed body
will be in equilibrium at all times.
Suppose that the body initially occupies a configuration C0 , in which a
particle X of the body occupies the position X, referred to a rectangular
Cartesian system (X1 , X2 , X3 ). Note that X is the name of the particle that
occupies the location X in configuration C0 , and therefore (X1 , X2 , X3 ) are
called the material coordinates. After the application of the loads, the body
moves and deforms and assumes a new configuration C. The particle X now
occupies the position x in the deformed configuration C (see Figure 1.1).
An analytical description of the deformation of a continuous body follows
one of the two approaches. In the first approach, called the material or
Lagrangian description, the motion of the body is referred to a reference
configuration CR , which is often chosen to be the undeformed configuration,
CR = C0 . Thus, in the Lagrangian description, the current coordinates
(x1 , x2 , x3 ) are expressed in terms of the reference coordinates (X1 , X2 , X3 ):
x = x(X, t) (1.1)

Figure 1.1: Reference and deformed configurations of a body.

2 THEORY OF ELASTICITY

and the variation of a typical variable φ over the body is described with respect
to the material coordinates (X1 , X2 , X3 ):
φ = φ(X, t) (1.2)
In the spatial or Eulerian description, the motion is referred to the current
configuration C occupied by the body, and φ is described with respect to the
position (x1 , x2 , x3 ) in space, currently occupied by material particle X:
φ = φ(x, t), X = X(x, t) (1.3)
The coordinates (x1 , x2 , x3 )are termed the spatial coordinates.
Equations (1.2) and (1.3) each convey a different information. In Eq. (1.2),
a change in time t implies that the same material particle X, occupying
position X in C0 , has a different value φ. Thus the attention is focused on
the material particle X. In Eq. (1.3), a change in time t implies that a
different value φ is observed at the same spatial location x, now probably
occupied by a different material particle X. Hence, attention is focused on a
spatial position x.

Example 1.1:
To illustrate the difference between the two descriptions further, consider the 1-D
mapping x = X(1 + 0.5t) defining the motion of a rod of initial length two units. The
rod experiences a temperature distribution T given by the material description T = 2Xt2 or
by the spatial description T = xt2 /(1 + 0.5t) (see Figure 1.2).
It is clear from Figure 1.2 that the particle material coordinate (label) X remains
associated with the particle while its spatial position x changes. The temperature at a given
time can be found in one of the two ways: for example, at time t = 3 the temperature
of the particle labeled X = 2 is T = 2 × 2(3)2 = 36; alternatively, the temperature
of the same particle which at t = 3 is at a spatial position x = 2(1 + 0.5 × 3) = 5 is
T = 2 × 5(3)2 /(1 + 0.5 × 3) = 36.

Figure 1.2: Material and spatial descriptions of motion.

 J. N. REDDY 3

In the study of solid bodies, the Eulerian description is less useful since
the configuration C is unknown. On the other hand, it is the preferred
description for the study of the motion of fluids because the configuration is
known and remains unchanged, and we wish to determine the changes in the
fluid velocities, pressure, density and so on. Thus, in the Eulerian description,
attention is focussed on a given region of space instead of on a given body of
matter. In much of the current study we focus our attention on the Lagrangian
description of the motion of solid bodies undergoing geometric changes.

2. Kinematics
2.1 Deformation Gradient Tensor
Consider two material particles P and Q in the neighborhood of each other in
the reference configuration C0 (see Figure 2.1). The positions of P and Q are
denoted by XP and XQ , respectively. The position of Q relative to P is given
by the elemental vector dX in C0 :
dX = XQ − XP
After deformation the material particles P and Q occupy spatial positions xP
and xQ , respectively in C; they are now labeled as P̄ and Q̄. The position of
Q̄ relative to P̄ is denoted by dx and it is given by
dx = xQ − xP

Figure 2.1: Deformation of a line segment P Q in a continuous medium.

4 THEORY OF ELASTICITY

The displacements of the material particles P and Q are given by

uP = xP − XP , uQ = xQ − XQ (2.1)

In Example 1.1, the displacement of a material point occupying position X in

C0 is
u(X, t) = x − X = X(1 + 0.5t) − X = 0.5Xt

One of the key quantities in finite deformation analysis is the deformation

gradient tensor F, which gives the relationship of a material line dX before
deformation to the line dx after deformation. It is defined as
³T
∂x
²
dx = F · dX = dX · FT where F = ≡ (∇o x)T (2.2)
∂X
and ∇0 is the gradient operator with respect to X. We also have
∂X
dX = F−1 · dx = dx · F−T , where F−T = ≡ ∇X (2.3)
∂x
and ∇ is the gradient operator with respect to x. In indicial notation, Eqs.
(2.2) and (2.3) can be written as
∂xi
F = FiI êi ÊI , FiI = (2.4a)
∂XI
−1 −1 ∂XI
F−1 = FIi ÊI êi , FIi = (2.4b)
∂xi
Here the lower case indices refer to the current (spatial) cartesian coordinates,
whereas upper case indices refer to the reference (material) Cartesian
coordinates. The determinant of F is called the Jacobian of the motion,
and it is denoted by J. The deformation tensor F can be expressed in terms
of the displacement vector as

F = (∇o x)T = (∇o u + I)T , FT = ∇o x = ∇o u + I (2.5)

Homogeneous deformation and Rigid Body Motion

The deformation tensor conveys no information about the translation of the
body. Further, if F=I everywhere in the body, then the body is not rotated
and is undeformed. If F has the same value at every material point in a body,
then the mapping x = x(X, t) is said to be a homogeneous motion of the
body. In general, at any given time t > 0, a mapping x = x(X, t) is said to
be a homogeneous motion if and only if it can be expressed as

x=A·X+c
 J. N. REDDY 5

where the second-order tensor A and vector c are constants; c represents a

rigid-body translation. Note that for a homogeneous motion we have F=A.

Example 2.1:
Consider the uniform deformation of a square of side two units and initially centered at
X = (0, 0). The deformation is defined by the mapping

x1 = 3.5 + X1 + 0.5X2 , x2 = 4 + X2 (inverse is: X1 = −1.5 + x1 − 0.5x2 , X2 = −4 + x2 )

which produces the deformed shape shown in Figure 9.2.2. The components of the
deformation gradient tensor and its inverse can be expressed in matrix form as
´ ∂x1 ∂x1 µ ´ µ ´ ∂X1 ∂X1 µ ´ µ
1.0 0.5 1.0 −0.5
[F ] = ∂X1
∂x2
∂X2
∂x2 = ; [F ]−1 = ∂x1
∂X2
∂x2
∂X2 =
∂X1 ∂X2
0.0 1.0 ∂x1 ∂x2
0.0 1.0

The displacement components (u1 , u2 ) are given by

u1 = x1 − X1 = 3.5 + 0.5X2 , u2 = x2 − X2 = 4.0

Ý 1 and E
The unit vectors E Ý 2 in the initial configuration deform to the vectors
´ µº » º » ´ µº » º »
1.0 0.5 1.0 1.0 1.0 0.5 0.0 0.5
= ; =
0.0 1.0 0.0 0.0 0.0 1.0 1.0 1.0

Ý1 and e
The unit vectors e Ý2 in the current configuration deformed from the vectors
´ µº » º » ´ µº » º »
1.0 −0.5 1.0 1.0 1.0 −0.5 0.0 −0.5
= ; =
0.0 1.0 0.0 0.0 0.0 1.0 1.0 1.0

Figure 2.2: Uniform deformation of a square.

6 THEORY OF ELASTICITY

2.2 Strain Tensors

Next we discuss a general measure of deformation, independent of both
translation and rotation. Consider two material particles P and Q in the
neighborhood of each other (separated by dX) in the reference configuration
(see Figure 2.1). In the deformed configuration the material points P and Q
are denoted by P̄ and Q̄, and they are separated by dx. We wish to determine
the change in the distance dX between the material points P and Q as the
body deforms and the material points move to the new locations P̄ and Q̄.
The distances between points P and Q and points P̄ and Q̄ are given,
respectively, by

(dS)2 = dX · dX (2.6a)
(ds)2 = dx · dx = (F · dX) · (F · dX) = dX · (FT · F) · dX
≡ dX · C · X (2.6b)

where C is the right Cauchy-Green deformation tensor

C = FT · F (2.7)

The change in the squared lengths that occurs as the body deforms from
the initial to the current configuration can be expressed relative to the original
length as
(ds)2 − (dS)2 = 2 dX · E · dX (2.8)

where E is called the Green-St. Venant (Lagrangian) strain tensor or

simply the Green strain tensor, which can be expressed as

1° T ± 1
E= F · F − I = (C − I) (2.9)
2 2
1h i
= (I + ∇o u)T · (I + ∇o u) − I
2
1h i
= (∇o u)T + ∇o u + (∇o u)T · (∇o u) (2.10)
2

Clearly, the Green strain tensor is symmetric. Also, the change in the squared
lengths is zero if and only if E = 0.
Alternatively, the change in the squared lengths that occurs as the body
deforms from the initial to the current configuration can be expressed relative
to the current length as

(ds)2 − (dS)2 = 2 dx · e · dx (2.11)

 J. N. REDDY 7

where e is called the Almansi-Hamel (Eulerian) strain tensor or simply

the Euler strain tensor, which can be expressed as

1° ± 1
e= I − F−T · F−1 = (I − c) (2.12)
2 2
1h T
i
= I − (I − ∇u) · (I − ∇u)
2
1h i
= (∇u)T + ∇u − (∇u)T · (∇u) (2.13)
2
where c is the Cauchy strain tensor, and its inverse is called the left Cauchy-
Green or Finger tensor.
In the Cartesian component form, we can write

E = EIJ ÊI ÊJ (2.14)

e = eij êi êj (2.15)

with components

1 ∂xm ∂xm 1 ∂uI ∂uJ ∂uK ∂uK

² ³ ² ³
EIJ = − δIJ = + + (2.16)
2 ∂XI ∂XJ 2 ∂XJ ∂XI ∂XI ∂XJ
À ! À !
1 ∂XK ∂XK 1 ∂ui ∂uj ∂uk ∂uk
eij = δij − = + − (2.17)
2 ∂xi ∂xj 2 ∂xj ∂xi ∂xi ∂Xj

In expanded notation, the Green strain components, for example, are given
by
"² ³2 ³2 ³2 #
∂u1 1 ∂u1 ∂u2 ∂u3
² ²
E11 = + + +
∂X1 2 ∂X1 ∂X1 ∂X1
"² ³2 ³2 ³2 #
∂u2 1 ∂u1 ∂u2 ∂u3
² ²
E22 = + + +
∂X2 2 ∂X2 ∂X2 ∂X2
"² ³ #
∂u3 1 ∂u1 2 ∂u2 2 ∂u3 2
³ ² ³ ²
E33 = + + +
∂X3 2 ∂X3 ∂X3 ∂X3
1 ∂u1 ∂u2 ∂u1 ∂u1 ∂u2 ∂u2 ∂u3 ∂u3
² ³
E12 = + + + +
2 ∂X2 ∂X1 ∂X1 ∂X2 ∂X1 ∂X2 ∂X1 ∂X2
1 ∂u1 ∂u3 ∂u1 ∂u1 ∂u2 ∂u2 ∂u3 ∂u3
² ³
E13 = + + + +
2 ∂X3 ∂X1 ∂X1 ∂X3 ∂X1 ∂X3 ∂X1 ∂X3
1 ∂u2 ∂u3 ∂u1 ∂u1 ∂u2 ∂u2 ∂u3 ∂u3
² ³
E23 = + + + + (2.18)
2 ∂X3 ∂X2 ∂X2 ∂X3 ∂X2 ∂X3 ∂X2 ∂X3
8 THEORY OF ELASTICITY

Example 2.2:
Consider a rectangular block of dimensions a × b × h, where h is very small compared to
a and b. Suppose that the block is deformed into the diamond shape shown in Figure
2.3. By inspection, the geometry of the deformed body can be described as follows: let
(X1 , X2 , X3 ) denote the coordinates of a material point in the undeformed configuration.
Thus the coordinate mapping and its inverse are given by
e ab ae0
x1 = X1 + 0 X2 , X1 = x1 − x2
b ab − e20 ab − e20
e0 be0 ab
x2 = X2 + X , X2 = − x1 + x2
a 1 ab − e20 ab − e20
x3 = X3 , X3 = x3 .
Thus, the displacement components of a material point in the Lagrangian description are
e e
u1 = x1 − X1 = 0 X2 , u2 = x2 − X2 = 0 X1 , u3 = x3 − X3 = 0.
b a
and the spatial description they are
e20 ae0
u1 = x1 − X1 = − x1 + x2
ab − e20 ab − e20
be0 e20
u2 = x2 − X2 = 2
x1 − x2
ab − e0 ab − e20
u3 = x3 − X3 = 0.
The only nonzero Green strain tensor components are given by
1 e0 2
±2
e e 1 e0
° ± °
E11 = , E12 = 0 + 0 , E22 = .
2 a 2b 2a 2 b
The Almansi strain tensor components are
e20 e20 (e20 + b2 )
´ µ
1
e11 = − − ,
ab − e20 2 (ab − e20 )2
e0 (a + b) e3 (a + b)
e12 = 2
+ 0 ,
(ab − e0 ) (ab − e20 )2
e20 e20 (e20 + a2 )
´ µ
1
e22 = − − .
ab − e20 2 (ab − e20 )2

Figure 2.3: Undeformed and deformed rectangular block.

 J. N. REDDY 9

The same results can be obtained using the elementary mechanics of materials approach,
where the strains are defined to be the ratio of the difference between the final length and
original length to the original length. For example, a line element AB in the undeformed
body moves to position ĀB̄. Then the strain in the line AB is given by
r ±2
ĀB̄ − AB 1 e0
q °
E11 = εAB = = a2 + e20 − 1 = 1+ −1
AB a a
´ ±2 µ ±2
1 e0 1 e0
° °
= 1+ + ··· − 1 ≈ .
2 a 2 a

Similarly, ´ µ
±2 ±2
1 e0 1 e0
° °
E22 = 1+ + ··· − 1 ≈ .
2 b 2 b

Example 2.3:

For the deformation given in Example 2.1 the right Cauchy–Green deformation tensor and
the Cauchy strain tensors are respectively
´ µ´ µ ´ µ
1.0 0.0 1.0 0.5 1.0 0.5
[C] = [F ]T [F ] = =
0.5 1.0 0.0 1.0 0.5 1.25
´ µ´ µ ´ µ
−T −1 1.0 0.0 1.0 −0.5 1.0 −0.5
[c] = [F ] [F ] = =
−0.5 1.0 0.0 1.0 −0.5 1.25

The Green and Almansi strain tensor components in matrix form are given by
´ µ ´ µ
1 0.0 0.5 1 0.0 0.5
[E] = ; [e] =
2 0.5 0.25 2 0.5 −0.25

Example 2.4:

Consider the uniform deformation of a square of side 2 units initially centered at X = (0, 0).
The deformation is given by the mapping

1 1
x1 = (18 + 4X1 + 6X2 ), x2 = (14 + 6X2 )
4 4

We wish to
(a) sketch the deformed configuration of the body,
(b) compute the components of the deformation gradient tensor F and its inverse (display
them in matrix form),
(c) compute the components of the right and left Cauchy-Green deformation tensors (display
them in matrix form), and
(d) compute Green’s and Almansi’s strain tensor components (display them in matrix form).
Solution:
(a) Sketches of the undeformed and deformed bodies are shown in Figure 2.4.
10 THEORY OF ELASTICITY

Figure 2.4: Undeformed and deformed rectangular block.

(b) Note that the inverse transformation is given by

º » ´ µ−1 ²º » º »³ º » ´ µº »
X1 4 6 x1 1 18 1 9 1 3 −3 x1
=4 − =− +
X2 0 6 x2 4 14 6 7 3 0 2 x2

The matrix form of the deformation gradient tensor and its inverse are
´ ∂x1 ∂x1 µ ´ µ ´ ∂X1 ∂X1 µ ´ µ
∂X1 ∂X2 1 2 3 −1 ∂x1 ∂x2 1 3 −3
[F ] = ∂x2 ∂x2 = ; [F ] = ∂X2 ∂X2 =
∂X1 ∂X2
2 0 3 ∂x1 ∂x2
3 0 2

(c) The right and left Cauchy–Green deformation tensors are, respectively,
´ µ ´ µ
1 2 3 1 13 9
[C] = [F ]T [F ] = , [c] = [F ][F ]T =
2 3 9 4 9 9

(d) The Green and Almansi strain tensor components in matrix form are respectively
´ µ
1  T ¡ 1 0 3
[E] = [F ] [F ] − [I] =
2 2 3 7
´ µ
1  1 0 9
[I] − [F ]−T [F ]−1 =
¡
[e] =
2 18 9 −4
 J. N. REDDY 11

2.3 Infinitesimal Strain and Rotation Tensors

When all displacements gradients are small (or infinitesmal), i.e.,
|∂uI /∂XJ | << 1, we can neglect the nonlinear terms (because they are even
smaller than |∂uI /∂XJ |) in Eq. (2.18). The infinitesimal strain tensor in
the Lagrangian coordinate system is denoted by ε,

1h i
ε= (∇u)T + ∇u (2.19)
2
In the case of infintesimal strains, no distinction is made between the material
coordinates X and the spatial coordinates x. Therefore, the linear Eulerian
strains and linear Lagrangian strains become the same. The linear strain-
displacement relations in the Cartesian system are given by

1
εij = (ui,j + uj,i ) (2.20a)
2
or, in expanded form,

∂u1 ∂u2 ∂u3

ε11 = ; ε22 = ; ε33 =
∂X1 ∂X2 ∂X3
1 ∂u1 ∂u2 1 ∂u1 ∂u3 1 ∂u2 ∂u3
² ³ ² ³ ² ³
ε12 = + ; ε13 = + ; ε23 = +
2 ∂X2 ∂X1 2 ∂X3 ∂X1 2 ∂X3 ∂X2
(2.20b)

The strains defined by Eqs. (2.10), (2.13) and (2.19) are valid in any
coordinate system. Hence, they can be expressed in component form in any
given coordinate system by exanding the strain tensors in the dyadic form
and the operator ∇ in that coordinate system. For example, in the cylindrical
coordinate system we have

u = ur êr + uθ êθ + uz êz , (2.21)

∂ 1 ∂ ∂
∇ = êr + êθ + êz . (2.22)
∂r r ∂θ ∂z
The only nonzero derivatives of the base vectors are

∂êr ∂êθ
= êθ , = −êr . (2.23)
∂θ ∂θ
Using u and ∇ from Eqs. (2.21) and (2.22), we obtain

∂ 1 ∂ ∂
² ³
∇u = êr + êθ + êz (ur êr + uθ êθ + uz êz )
∂r r ∂θ ∂z
12 THEORY OF ELASTICITY

∂ur ∂uθ ∂uz 1 ∂ur

= êr êr + êr êθ + êr êz + êθ êr
∂r ∂r ∂r r ∂θ
ur ∂êr 1 ∂uθ uθ ∂êθ 1 ∂uz
+ êθ + êθ êθ + êθ + êθ êz
r ∂θ r ∂θ r ∂θ r ∂θ
∂ur ∂uθ ∂uz
+ êz êr + êz êθ + êz êz
∂z ∂z ∂z
∂ur ∂uθ 1 ∂ur uθ
= êr êr + êr êθ + êθ êr − êθ êr
∂r ∂r r ∂θ r
∂uz ∂ur ur 1 ∂uθ
+ êr êz + êz êr + êθ êθ + êθ êθ
∂r ∂z r r ∂θ
1 ∂uz ∂uθ ∂uz
+ êθ êz + êz êθ + êz êz
r ∂θ ∂z ∂z
∂ur ∂uθ 1 ∂u r uθ
(∇u)T = êr êr + êθ êr + êr êθ − êr êθ
∂r ∂r r ∂θ r
∂uz ∂ur ur 1 ∂uθ
+ êz êr + êr êz + êθ êθ + êθ êθ
∂r ∂z r r ∂θ
1 ∂uz ∂uθ ∂uz
+ êz êθ + êθ êz + êz êz
r ∂θ ∂z ∂z
Substituting the above expressions into Eq. (2.19) and collecting the
coefficients of various dyads (i.e., coefficients of êr êr , êr êθ , etc.) we obtain
the linear Lagrange strain components
∂ur 1 1 ∂ur ∂uθ uθ 1 ∂ur ∂uz
² ³ ² ³
εrr = , εrθ = + − , εrz = + ,
∂r 2 r ∂θ ∂r r 2 ∂z ∂r
ur 1 ∂uθ 1 ∂uθ 1 ∂uz ∂uz
² ³
εθθ = + , εzθ = + , εzz = . (2.24)
r r ∂θ 2 ∂z r ∂θ ∂z

The displacement gradient tensor ∇u (note that F is the deformation

gradient tensor) can be expressed as the sum of a symmetric tensor and
antisymmetric tensor
1h i 1h i
∇u = (∇u)T + ∇u + ∇u − (∇u)T ≡ ε + ω (2.25)
2 2
where the symmetric part is clearly the infinitesimal strain tensor, and the
antisymmetric part is known as the infinitesimal rotation tensor
1h i
ω= ∇u − (∇u)T (2.26)
2
It is clear that ω is antisymmeric or skew-symmetric, i.e., ω T = −ω. In
Cartesian component form,

ωij = (ui,j − uj,i ) , ωij = −ωji (2.27)

 J. N. REDDY 13

Thus, there are only three independent components of ω:

−ω12 −ω13
 
0
1
[ω] =  ω12 0 −ω23  (2.28)
2
ω13 ω23 0

While there is no restriction placed on the magnitude of ∇u in writing (2.25), ε

and ω do not have the meaning of infinitesimal strain and infinitesimal rotation
tensors unless the deformation is infinitesimal (i.e., ∇u is small).
Since ω has only three independent components, three components can be
used to define the components of a vector θ

1
ωij = −eijk θk , or θk = − ekij ωij (2.29)
2
It follows that
1 1
θk = ekij uj,i or θ = curl u = ∇ × u (2.30)
2 2

Certain motions do not produce infinitesimal strains but they may produce
finite strains. For example, the following displacement field

u1 = b1 + c2 X3 − c3 X2
u2 = b2 + c3 X1 − c1 X3
u3 = b3 + c1 X2 − c2 X1

for arbitrary constants (bi , ci ) produces no infinitesimal strains, but the fnite
strain components due to the displacement field are not zero. The deformation
gradient tensor components associated with the dispacement field are

1 −c3
 
c2
[F ] =  c3 1 −c1 
−c2 c1 1

Thus, [F ] = [I] if and only if all ci = 0, and for any sufficiently large ci 6= 0,
the mapping does not represent a rigid body motion. Figure 2.5 depicts the
deformation for the two-dimensional case, with b1 = 2, b2 = 3, and c3 = 1.
Clearly, the motion is not a rigid body motion, although the infinitesimal
strains produced by the motion are zero. The finite strains and deformation
gradient tensor give true measures of the deformation.
14 THEORY OF ELASTICITY

Figure 2.5: A mapping that produces zero infinitesimal strains but nonzero
finite strains.

2.4 Polar Decomposition

Recall that the deformation gradient tensor F transforms a material vector
dX into the corresponding spatial vector dx, and it forms an essential part of
the definition of any strain measure. Another role of F in connection with the
strain measures is discussed with the help of the polar decomposition theorem
of Cauchy, which enables F to be written as

F=R·U (2.31)

where R is an orthogonal rotation tensor and U is a symmetric stretch tensor

(stretch is the ratio of the final length to the original length). An orthogonal
rotation tensor R is one that satisfies RT · R = I.
To evaluate the tensors R and U, we recall the definition of C:

C = FT · F = UT · RT · R · U = UT · U (2.32)

To use Eq. (2.32) to obtain U, it is necessary to write C in terms of its

eigenvalues and principal directions. Let

3
X
C= λ2α N̂α N̂α (2.33)
α=1
 J. N. REDDY 15

where λ2α are the eigenvalues and N̂α are the eigenvectors of C. Then it follows
that
3
X
U= λα N̂α N̂α (2.34)
α=1

Once the stretch tensor is known, the rotation tensor R can be obatined from
Eq. (2.31) as R = F · U−1 .

Example 2.5:
Consider the deformation given by the mapping

1 1
x1 = [4X1 + (9 − 3X1 − 5X2 − X1 X2 ) t] , x2 = [4X2 + (16 + 8X1 ) t]
4 4

(a) For X = (0, 0) and t = 1 determine the deformation gradient tensor F and right Cauchy-
Green tensor C.
(b) Find the eigenvalues (stretches) λ1 and λ2 and the associated eigenvectors N1 and N2 .
(c) Use the polar decomposition to determine the symmetric stretch tensor U and rotation
tensor R.
Solution:

(a) For X = (0, 0) and time t = 1 the components of the deformation gradient tensor F and
right Cauchy–Green tensor C are
´ µ ´ µ
1 1 −5 1 65 27
[F ] = , [C] =
4 8 4 16 27 41

(b) The eigenvalues λ21 and λ22 of matrix [C] are determined by setting

|[C] − λ2 [I]| = 0 → λ21 = 5.1593, λ22 = 1.4658

so that λ1 = 2.2714 and λ2 = 1.2107. The eigenvectors are (in vector component form)
º » º »
0.8385 −0.5449
{N (1) } = , {N (2) } =
0.5449 0.8385

(b) Hence, the stretch tensor can be written as

Ý (1) N
U = λ1 N Ý (1) + λ2 N
Ý (2) N
Ý (2)
° ±° ±
(1)
= λ1 N1 e Ý1 + N2(1) e
Ý2 (1)
N1 e Ý1 + N2(1) e
Ý2
° ±° ±
(2)
+ λ2 N1 e Ý1 + N2(2) e
Ý2 (2)
N1 e Ý1 + N2(2) e
Ý2
° ± ° ±
(1) (2)
= λ1 [N1 ]2 + λ2 [N1 ]2 e Ý1 + λ1 [N2(1) ]2 + λ2 [N2(2) ]2 e
Ý1 e Ý2 e
Ý2
° ±
+ λ1 N1(1) N2(1) + λ2 N1(2) N2(2) (Ý Ý2 + e
e1 e Ý2 e
Ý1 )
16 THEORY OF ELASTICITY

or in matrix form ´ µ
1.9564 0.4846
[U ] =
0.4846 1.5257

Then the rotation tensor [R] in matrix form is given by

´ µ
0.3590 −0.9333
[R] = [F ][U ]−1 =
0.9333 0.3590

3. Compatibility Equations
The task of computing strains (infinitesimal or finite) from a given
displacement field is a straightforward exercise. However, the determination
of displacements from a given strain field is not as straightforward due to
the fact there are six independent partial differential equations (i.e., strain-
displacement relations) for only three unknown displacements, which would
in general overdetermine the solution. We will find some conditions, known
as St. Venant’s compatibility equations. The derivation is presented for
infinitesimal strains. For finite strains the same steps may be followed but
the process is so difficult that it is never attempted (although some general
compatibility conditions may be stated to insure integrability of the six
nonlinear partial differential equations).
To understand the meaning of strain compatibility, imagine that a material
body is cut up into pieces before it is strained, and then each piece is
given a certain strain. The strained pieces cannot be fitted back into a
single continuous body without further deformation. On the other hand if
the strain in each piece is related to or compatible with the strains in the
neighboring pieces, then they can be fitted together to form a continuous
body. Mathematically, the six relations that connect six strain components to
the three displacement components should be consistent. To make this point
clear, consider the two-dimensional case. We have three strain-displacement
relations in two displacements:

∂u1
= ε11 (3.1)
∂x1
∂u2
= ε22 (3.2)
∂x2
∂u1 ∂u2
+ = 2ε12 (3.3)
∂x2 ∂x1

If the given data (ε11 , ε22 , ε12 ) is compatible (or consistent), any two of
the three equations should yield the same displacement components. The
compatibility of the data can be established as follows. Differentiate the first
 J. N. REDDY 17

equation with respect to x2 twice, the second equation with respect to x1

twice, and the third equation with respect to x1 and x2 each, we obtain

∂ 3 u1 ∂ 2 ε11
= (3.10 )
∂x1 ∂x22 ∂x22
∂ 3 u2 ∂ 2 ε22
2 = (3.20 )
∂x2 ∂x1 ∂x21
∂ 3 u1 ∂ 3 u2 ∂ 2 ε12
+ = 2 (3.30 )
∂x22 ∂x1 ∂x21 ∂x2 ∂x1 ∂x2

Using Eqs. (3.1’) and (3.2’) in (3.3’), we arrive at the following equation
between the three strains:

∂ 2 ε11 ∂ 2 ε22 ∂ 2 ε12

2 + 2 =2 (3.4)
∂x2 ∂x1 ∂x1 ∂x2

Equation (2.35) is called the strain compatibility condition among the three
strains for a 2D case.
Similar procedure can be followed to obtain the strain compatibility
conditions for the 3D case. In addition to Eq. (3.4) five more such conditions
can be derived:
∂ 2 ε11 ∂ 2 ε33 ∂ 2 ε13
+ = 2 (3.5)
∂x23 ∂x21 ∂x1 ∂x3

∂ 2 ε22 ∂ 2 ε33 ∂ 2 ε23

+ = 2 (3.6)
∂x23 ∂x22 ∂x2 ∂x3

∂ 2 ε11 ∂ 2 ε23 ∂ 2 ε13 ∂ 2 ε12

+ 2 = + (3.7)
∂x2 ∂x3 ∂x1 ∂x1 ∂x2 ∂x1 ∂x3

∂ 2 ε22 ∂ 2 ε13 ∂ 2 ε23 ∂ 2 ε12

+ = + (3.8)
∂x1 ∂x3 ∂x22 ∂x1 ∂x2 ∂x2 ∂x3

∂ 2 ε33 ∂ 2 ε12 ∂ 2 ε13 ∂ 2 ε23

+ = + (3.9)
∂x1 ∂x2 ∂x23 ∂x2 ∂x3 ∂x1 ∂x3
The six equations can be written index notation as

∂ 2 εmn ∂ 2 εij ∂ 2 εim ∂ 2 εjn

+ = + (3.10)
∂xi ∂xj ∂xm ∂xn ∂xj ∂xn ∂xi ∂xm

These are both necessary and sufficient to determine a single-valued

displacement field.
18 THEORY OF ELASTICITY

Equation (3.10) can be derived in vector form as follows. We begin with

the curl of ε:
À !
∂εjr 1 ∂ 2 uj ∂ 2 ur
∇ × ε = eijk êk êr = eijk + êk êr
∂xi 2 ∂xi ∂xr ∂xi ∂xj
À !
1 ∂ 2 uj
= eijk + 0 êk êr (3.11)
2 ∂xi ∂xr

Using Eq. (2.30) we have

1 ∂ 2 uj ∂ 1 ∂uj ∂θk
² ³
∇ × ε = eijk êk êr = eijk êk êr = êk êr
2 ∂xi ∂xr ∂xr 2 ∂xi ∂xr
or
∂θk ∂
(∇ × ε)T = êr êk = êr (θk êk ) = ∇θ
∂xr ∂xr
Since curl of a the gradient of a vector (or tensor) is zero, we take curl of the
above equation and arrive at the compatibility equation in vector/tensor form:

∇ × (∇ × ε)T = 0 or eikr ejls εij,kl = 0 (3.12)

Example 3.1:
Consider the problem of an isotropic cantilever beam bent by a load F0 at the free end (see
Figure 3.1). From the elementary beam theory, we have the following infinitesimal strains

F0 x1 x2 F0 x1 x2 (1 + ν)F0 2
ε11 = − , ε22 = −νε11 = ν , ε12 = − (h − x22 ), (3.13)
EI3 EI3 2EI3

where ν is the Poisson ratio, EI3 is the bending stiffness (E is Young’s modulus and I3 is
the second moment of area about the x3 −axis), and 2h is the height of the beam. We wish
to determine the two-dimensional displacement field (u1 , u2 ) in the beam.

Figure 3.1: Cantilever beam bent by a point load, F0 .

 J. N. REDDY 19

First, we determine if the given strains are compatible. Substituting εij into the single
compatibility equation for 2D, we obtain 0 + 0 = 0. Thus the strains satisfy the compatibility
equation in two dimensions. Although the two-dimensional strains are compatible, the
three-dimensional strains are not compatible. For example, using the additional strains,
ε33 = −νε11 , ε13 = ε23 = 0, one can show that all of the equations except the one in Eq.
(3.9) are satisfied.
Since the strains are compatible in two-dimensional theory of elasticity, we wish to
compute the displacement field u = e Ý2 u2 of the beam using the known strains.
Ý1 u1 + e
Integrating the strains in Eq. (3.13), we obtain

∂u1 F x x F0 x21 x2
ε11 = =− 0 1 2 or u1 = + f1 (x2 ), (3.14a)
∂x1 EI3 2EI3
∂u2 νF0 x1 x2 νF0 x1 x22
ε22 = = or u2 = + f2 (x1 ), (3.14b)
∂x2 EI3 2EI3

where (f1 , f2 ) are functions of integration. Substituting u1 and u2 into the definition of the
shear strain 2ε12 , we obtain

∂u1 ∂u2 F0 x21 df νF0 x22 df

2ε12 = + =− + 1 + + 2
∂x2 ∂x1 2EI3 dx2 2EI3 dx1

But this must be equal to the strain value given in Eq. (3.13):

F0 x21 df νF0 x22 df (1 + ν)

− + 1 + + 2 =− F0 (h2 − x22 ). (3.15)
2EI3 dx2 2EI3 dx1 EI3

Separating the x1 and x2 terms, we obtain

df2 F0 2 (1 + ν)F0 h2 df (2 + ν)F0 2

− + x − = 1 − x2 .
dx1 2EI3 1 EI3 dx2 2EI3

Since the left side depends only on x1 and the right side depends only on x2 , and yet the
equality must hold, it follows that both sides should be equal to a constant, say c0 :

df1 (2 + ν)F0 2 df2 F0 2 (1 + ν)F0 h2

− x2 = c0 , − + x − = c0 .
dx2 2EI3 dx1 2EI3 1 EI3

Integrating the expressions for f1 and f2 , we obtain

(2 + ν)F0 3
f1 (x2 ) = x2 + c0 x2 + c1
6EI3
F0 3 (1 + ν)F0 h2
f2 (x1 ) = x − x1 − c0 x1 + c2 ,
6EI3 1 EI3

where c1 and c2 are constants of integration that are to be determined. The displacements
(u1 , u2 ) are now given by

F0 2 (2 + ν)F0 3
u1 (x1 , x2 ) = − x x + x2 + c0 x2 + c1 , (3.16a)
2EI3 1 2 6EI3
(1 + ν)F0 h2 νF0 F0 3
u2 (x1 , x2 ) = − x1 + x x2 + x − c0 x1 + c2 . (3.16b)
EI3 2EI3 1 2 6EI3 1
20 THEORY OF ELASTICITY

The constants c0 , c1 , and c2 can be evaluated using the boundary conditions on the
displacements. We impose the following boundary conditions:
¬
∂u1 ¬
° ±¬
u1 (L, 0) = 0, u2 (L, 0) = 0, = 0. (3.17)
∂x2 ¬
¬
(L,0)

Substituting the expressions for u1 and u2 into the boundary conditions (3.17), we obtain

u1 (L, 0) = 0 → c1 = 0
F0 L3 (1 + ν)F0 h2 L
u2 (L, 0) = 0 → − c0 L + c2 = ,
6EI3 EI3
¬
∂u1 ¬ F0 L2 F0 L2
° ±¬
=0 → − +c=0 or c= .
∂x2 ¬ 2EI3 2EI3
¬
(L,0)

Finally, the displacement field of two-dimensional elasticity theory becomes

F0 (2 + ν)F0 3
L2 − x21 x2 +
  ¡
u1 (x1 , x2 ) = x2 , (3.18a)
2EI3 6EI3
F0 ¡ F (1 + ν)h2
x1 − 3L2 x1 + 2L3 + 0
  3
u2 (x1 , x2 ) = (L − x1 )
6EI3 EI3
νF0
+ x x2 . (3.18b)
2EI3 1 2

The values of the constants c, c1 , and c2 depend on the boundary conditions used. For
example, if one uses (∂u2 /∂x1 ) = 0 in place of (∂u1 /∂x2 ) = 0 at x1 = L and x2 = 0, the
resulting displacements would be

F0 (2 + ν)F0   2
L2 − x21 x2 − x2 3h − x22 ,
  ¡ ¡
u1 (x1 , x2 ) = (3.19a)
2EI3 6EI3
F0   3 ¡ νF0
u2 (x1 , x2 ) = x1 − 3L2 x1 + 2L3 + x x2 . (3.19b)
6EI3 2EI3 1 2

The displacement field in the beam according to the Euler–Bernoulli beam theory is
given by

F0
L2 − x21 x2 ,
  ¡
u1 (x1 , x2 ) = (3.20a)
2EI3
F0
x31 − 3L2 x1 + 2L3 ,
  ¡
u2 (x1 , x2 ) = (3.20b)
6EI3

and according to the Timoshenko beam theory, it is

F0
L2 − x21 x2 ,
  ¡
u1 (x1 , x2 ) = (3.21a)
2EI3
F0 F0
x31 − 3L2 x1 + 2L3 +
  ¡
u2 (x1 , x2 ) = (L − x1 ), (3.21b)
6EI3 Ks GA

where Ks is the shear correction factor, and A is the area of cross section. The transverse
deflection u2 of the Timoshenko beam theory has an extra term due to the inclusion of the
 J. N. REDDY 21

transverse shear force/strain effect. For a rectangular beam of height 2h and width b, we
have A = 2bh and I3 = 2bh3 /3.
A comparison of the elasticity solution (3.19a,b) against the Timoshenko beam solution
(3.21a,b) indicates that the elasticity solution has some higher-order terms that are negligible
(x2 ∼ h). Also one can see that (GA = 3EI3 /2[(1 + ν)h2 ]) the shear correction factor is
Ks = 2/3.

4. Measures of Stress
4.1 Cauchy Stress Tensor and Cauchy’s Formula
The equations of motion or equilibrium of a material body must be derived
for the deformed configuration of the body at time t. However, since the
geometry of the deformed configuration is not known, the equations must
be written in terms of the known reference configuration. In doing so we
introduce various measures of stress. They emerge in a natural way as we
transform volumes and areas from the deformed configuration to undeformed
(or reference) configuration.
First we introduce the true stress, i.e., stress in the deformed configuration.
Consider a deformed body at its current position. The surface force acting
on a small element of area in a continuous medium depends not only on the
magnitude of the area but also upon the orientation of the area. It is customary
to denote the direction of a plane area by means of a unit vector drawn normal
to that plane. To fix the direction of the normal, we assign a sense of travel
along the contour of the boundary of the plane area in question. The direction
of the normal is taken by convention as that in which a right-handed screw
advances as it is rotated according to the sense of travel along the boundary
curve or contour. Let the unit normal vector be given by n Ý . Then the area
can be denoted by s = sÝ n.
If we denote by df(n̂) the force on a small area n̂da located at the position
x, the stress vector can be defined as (see Figure 4.1)

∆f(n̂)
t(n̂) = lim . (4.1)
∆a→0 ∆a

We see that the stress vector is a point function of the unit normal n̂ which
denotes the orientation of the surface ∆a. The component of t that is in the
direction of n̂ is called the normal stress. The component of t that is normal
to n̂ is called a shear stress. Because of Newton’s third law for action and
reaction, we see that t(−n̂) = −t(n̂).
At a fixed point r for each given unit vector n̂ there is a stress vector t(n̂)
acting on the plane normal to n̂. Note that t(n̂) is, in general, not in the
direction of n̂. It is fruitful to establish a relationship between t and n̂.
22 THEORY OF ELASTICITY

Figure 4.1: Stress vector on a plane normal to n̂.

To estblish the relationship between t and n̂ we now set up an infinitesimal

tetrahedron in Cartesian coordinates as shown in Figure 4.2. If −t1 , −t2 , −t3 ,
and t denote the stress vectors in the outward directions on the faces of the
infinitesimal tetrahedron whose areas are ∆a1 , ∆a2 , ∆a3 , and ∆a, respectively,
we have by Newton’s second law for the mass inside the tetrahedron,
t∆a − t1 ∆a1 − t2 ∆a2 − t3 ∆a3 + ρ∆vf = ρ∆va (4.2)
where ∆V is the volume of the tetrahedron, ρ the density, f the body force
per unit mass, and a the acceleration. Since the total vector area of a closed
surface is zero (use the gradient theorem), we have
∆a n̂ − ∆a1 ê1 − ∆a2 ê2 − ∆a3 ê3 = 0. (4.3)

Figure 4.2: Tetrahedral element in Cartesian coordinates.

 J. N. REDDY 23

It follows that

∆a1 = (n̂ · ê1 )∆a, ∆a2 = (n̂ · ê2 )∆a, ∆a3 = (n̂ · ê3 )∆a. (4.4)

The volume of the element ∆v can be expressed as

∆h
∆v = ∆a, (4.5)
3
where ∆h is the perpendicular distance from the origin to the slant face.
Substitution of Eqs. (4.3) and (4.4) in (4.2) and dividing throughout by
∆S reduces it to
∆h
t = (n̂ · ê1 )t1 + (n̂ · ê2 )t2 + (n̂ · ê3 )t3 + ρ (a − f). (4.6)
3
In the limit when the tetrahedron shrinks to a point, ∆h → 0, we are left with

t = (n̂ · ê1 )t1 + (n̂ · ê2 )t2 + (n̂ · ê3 )t3

= (n̂ · êi )ti . (4.7)

It is now convenient to display the above equation as

t = n̂ · (ê1 t1 + ê2 t2 + ê3 t3 ) . (4.8)

The terms in the parenthesis are to be treated as a dyadic, called stress dyadic
↔
or stress tensor σ :
↔
σ ≡ ê1 t1 + ê2 t2 + ê3 t3 . (4.9)
The stress tensor is a property of the medium that is independent of the n̂.
Thus, we have
↔ ↔T
t(n̂) = n̂ · σ = σ · n̂ (4.10)
and the dependence of t on n̂ has been explicitly displayed. In Cartesian
component form, we have ti = nj σji .
It is useful to resolve the stress vectors t1 , t2 , and t3 into their orthogonal
components. We have

ti = σi1 ê1 + σi2 ê2 + σi3 ê3

= σij êj (4.11)

for i = 1, 2, 3. Hence, the stress tensor can be expressed in summation notation

as
↔
σ = êi ti = σij êi êj . (4.12)
24 THEORY OF ELASTICITY

The component σij represents the stress (force per unit area) on an area
perpendicular to the ith coordinate and in the jth coordinate direction (see
Figure 4.3). The stress vector t represents the vectorial stress on an area
perpendicular to the direction n̂. Equation (4.10) is known as the Cauchy
↔
stress formula, and σ is termed the Cauchy stress tensor. Thus, the Cauchy
stress tensor σ is defined to be the current force per unit deformed area
df = tda = σ · da, where Cauchy’s formula, t = n̂ · σ = σ T · n̂, is used.
The matrix form of the Cauchy’s formula (for computing purposes) is given
by     
 t1  σ11 σ21 σ31  n1 
t =  σ12 σ22 σ32  n2 (4.13)
 2
t3 σ13 σ23 σ33 n3
 

The component of the stress vector t normal to the plane n̂ is given by

tnn = t · n = ti ni = nj σji ni (4.14)

and the the component of t in the plane n̂ is

q
tns = |t|2 − t2nn (4.15)

Example 4.1:
With reference to a rectangular Cartesian system (x1 , x2 , x3 ), the components of the stress
dyadic at a certain point of a continuous medium are given by
"
200 400 300
#
[σ] = 400 0 0 psi.
300 0 −100

Figure 4.3: Display of stress components in Cartesian rectangular coordinates.

 J. N. REDDY 25

We wish to determine the stress vector t at the point and normal to the plane, x1 + 2x2 +
2x3 − 6 = 0, and then compute the normal and tangential components of the stress vector
at the point.
First we should find the unit normal to the plane on which we are required to find the
stress vector. The unit normal to a surface defined by the equation φ(x1 , x2 , x3 ) =constant
is given by
∇φ
Ý=
n , φ(x1 , x2 , x3 ) = x1 + 2x2 + 2x3 − 6,
|∇φ|
1
Ý = (Ý
n e + 2Ý e2 + 2Ý
e3 ).
3 1
The components of the stress vector are
(
t1
) "
200 400 300
# ( )
1
(
1600
)
1 1
t2 = 400 0 0 2 = 400 psi,
3 3
t3 300 0 −100 2 100
or
1
t(Ý
n) = (1600Ýe1 + 400Ý
e2 + 100Ýe3 ) psi.
3
The normal component tnn of the stress vector t on the plane is given by
2600
tnn = t(Ý Ý=
n) · n psi,
9
and the tangential component is given by (the Pythagorean theorem)
p 102 p
tns = |t|2 − t2nn = (256 + 16 + 1)9 − 26 × 26 psi
9
√
1781
= 100 = 468.9 psi.
9

4.2 Other Stress Measures

First Piola–Kirchhoff stress tensor
To express the elemental force df in terms of a stress times the initial
undeformed area dA requires a new stress tensor P,

df = t da = n̂ · σ da = da · σ where da = da n̂ (4.16)
= dA · P, where dA = dA N̂ (4.17)

where N̂ is the unit normal to the undeformed area dA. The stress tensor P is
called the first Piola–Kirchhoff stress tensor, and it gives the current force
per unit undeformed area. The tensor Cartesian component representation
of P is given by
P = PIi ÊI êi (4.18)
Thus the first Piola–Kirchhoff stress tensor is a mixed tensor and it is
unsymmetric.
26 THEORY OF ELASTICITY

Clearly, the first Piola–Kirchhoff stress tensor is related to the Cauchy

stress tensor [see Eqs. (4.16) and (4.17)]:

da · σ = dA · P

Using the Nanson’s formula

da = n̂da = F−T · N̂ JdA = N̂ · F−1 JdA = JdA · F−1 (4.19)

where J is the determinant of F, we can write

° ±
dA · P = da · σ = J dA · F−1 · σ

from which we have

P = JF−1 · σ (4.20)

Second Piola–Kirchhoff stress tensor

The second Piola–Kirchhoff stress tensor S, which is used in the
study of of large deformation analysis, is introduced as follows. Recall that
dX = F−1 · dx. Analogously, we can transform the force df on the deformed
elemental area da to the force dF on the undeformed elemental area dA (not
to be confused between the elemental force dF and deformation tensor F)

dF = F−1 · df = F−1 · (dA · P) = dA · P · F−T

≡ dA · S (4.21)

Thus, the second Piola–Kirchhoff stress tensor gives the transformed current
force per unit undeformed area. Cartesian component representation of S is

S = SIJ ÊI ÊJ (4.22)

The second Piola–Kirchhoff stress tensor is symmetric whenever the Cauchy

stress tensor is symmetric.
The Cauchy stress tensor, the second Piola–Kirchhoff stress tensor S, and
the first Piola–Kirchhoff stress tensor are connected to each other according
to the relations [see Eqs. (4.20) and (4.21)]

P = J F−1 · σ = S · FT (4.23)
1 1
σ = F · P = F · S · FT (4.24)
J J
S = J F−T · σ · F−1 = P · F−T (4.25)
 J. N. REDDY 27

5. Material Time Derivative and

Conservation of Mass
5.1 Material Derivative in Spatial Description
The partial time derivative with the material coordinates X held constant
should be distinguished from the partial time derivative with spatial
coordinates x held constant due to the difference in the descriptions of motion.
The material time derivative, D/Dt, is the time derivative with the material
coordinates held constant. The time derivative of a function φ in material
description (i.e., φ = φ(X, t)) with X held constant is nothing but the partial
derivative with respect to time:

Dφ ∂φ ∂φ
² ³
≡ = (5.1)
Dt ∂t X=const ∂t

In particular we have

∂x ∂x
² ³ ² ³
= ≡v (5.2)
∂t X=const ∂t X=const

where v is the velocity vector. Similarly the acceleration is

∂v ∂v
² ³ ² ³
= ≡a (5.3)
∂t X=const ∂t X=const

where a is the acceleration vector.

° ±In spatial description, for example
° ± v = v(x, t), the partial time
° derivative
±
∂v ∂v ∂v
∂t X=const is different from ∂t x=const . The time derivative ∂t X=const
denotes the local rate of change of v; it is the rate of change of velocity read
by a velocity meter located at the fixed spatial location x, which is not the
same as the acceleration of the particle just passing the place x.
To calculate the material time derivative of a function of spatial
coordinates, we assume that there exists differentiable mapping x = x(X, t)
so that we can write φ(x, t) = φ[x(X, t), t] and compute the derivative using
chain rule of differentiation:

Dφ ∂φ ∂φ ∂xi ∂φ
² ³ ² ³ ² ³
≡ = +
Dt ∂t X=const ∂t x=const ∂t X=const ∂xi
∂φ ∂φ
² ³
= + vi
∂t x=const ∂xi
∂φ
² ³
= + v · ∇φ
∂t x=const
28 THEORY OF ELASTICITY

where Eq. (5.2) is used in the second line. Thus, the material derivative
operator is given by
D ∂
² ³
= +v·∇ (5.4)
Dt ∂t x=const

Example 5.1:
This example illustrates the calculation of material time derivative based on material spatial
descriptions. Suppose that the motion is described by the mapping, x = (1 + t)X. Then the
velocity in the material and spatial descriptions respectively are

∂x x
v(X) = =X or v(x) = (a)
∂t 1+t

Now consider a function φ(X, t) = Xt2 in material description. The material time
derivative of φ in the material description is simply

∂φ
= 2Xt
∂t

The material time derivative of φ(x, t) = xt2 /(1 + t) is

Dφ ∂φ ∂φ
= +v
Dt ∂t
² ∂x ³
±² 2 ³
2xt + xt2 x t
°
= +
(1 + t)2 1+t 1+t
2xt
= (b)
1+t

which is the same as that in Eq. (a), except that it is expressed in terms of the current
coordinate, x.

The material time derivative of the volume integral of a function φ(x, t)

measured per unit mass over a spatial volume v bounded by s is given by

D Dφ
Z Z
ρφ dv = ρ dv (5.5)
Dt v v Dt

Equation (5.5) is known as the Reynolds transport theorem.

5.2 Continuity Equation in Material Description

Under the assumption that the mass is neither created nor destroyed during the
motion, we require that the total mass of any material volume be the same at
any instant during the motion. To express this in analytical terms, we consider
a material body that occupies region/configuration C0 with density ρ0 and
volume V at time t = 0. The same material body occupies the configurations
 J. N. REDDY 29

C at time t > 0, and it has a density ρ and vlume v. As per the principle of
conservation of mass, we have
Z Z
ρ0 dV = ρdv (5.6)
V v

Using the relation between dV and dv, dv = J dV we arrive at

Z
(ρ0 − Jρ) dV = 0 (5.7)
V

This is the global form of the continuity equation. Since the material volume
V we selected is arbitrarily small, as we shrink the volume to a point, we obtain
the local form of the continuity equation

ρ0 = Jρ (5.8)

5.3 Continuity Equation in Spatial Description

Now we consider an arbitrary volume v fixed in space, bounded by surface
s. Conservation of mass in this case means that the time rate of change of
of mass in v is equal to the mass inflow through the surface. If a continuous
medium of density ρ fills the volume at time t, the toal mass in v is
Z
M= ρ(x, t) dv
v

The rate of increase of mass in V is

∂M ∂ρ
Z
= dv (5.9)
∂t v ∂t

The mass inflow across the surface is

I I
(−ρvn ) ds = − ρv · n̂ ds
s Zs
=− ∇ · (ρv)dv (5.10)
v

where the divergence theorem is used in arriving at the last line. Equating
Eqs. (5.9) and (5.10), we obtain

∂ρ
Z Z
dv = − ∇ · (ρv)dv
v ∂t v
30 THEORY OF ELASTICITY

or
∂ρ
Z ´ µ
+ ∇ · (ρv) dv = 0 (5.11)
v ∂t
which is the global form of the continuity equation in spatial description. The
local form is given by
∂ρ
+ ∇ · (ρv) = 0 (5.12)
∂t
Alternate form of Eq. (5.12) is obtained as follows:

∂ρ
0= + ∇ · (ρv)
∂t
∂ρ
= + v · ∇ρ + ρ∇ · v
∂t
Dρ
= + ρ∇ · v (5.13)
Dt

where the definition of material time derivative, Eq. (5.4), is used in arriving
at the last line.

6. Conservation of Linear and Angular Momenta

6.1 Equations of Motion
The principle of conservation of linear momentum, or Newton’s second law of
motion, applied to a set of particles (or rigid body) can be stated as the time
rate of change of (linear) momentum of a collection of particles equals the net
force exerted on the collection. Written in vector form, the principle implies

d
(mv) = F, (6.1)
dt
where m is the total mass, v the velocity, and F the resultant force on the
collection particles. For constant mass, Eq. (6.1) becomes

dv
F=m = ma, (6.2)
dt
which is the familiar form of Newton’s second law. To derive the equation of
motion applied to a fixed region in space through which material flows, we
must identify the forces acting on it.
Forces acting on a volume element can be classified as internal and
external. The internal forces resist the tendency of one part of the region/body
to be separated from another part. The internal force per unit area is termed
stress, as defined in Section 4. The external forces are those transmitted by
 J. N. REDDY 31

the body. The external forces can be further classified as body (or volume)
forces and surface forces.
Body forces act on the distribution of mass inside the body. Examples
of body forces are provided by the gravitational and electromagnetic forces.
Body forces are usually measured per unit mass or unit volume of the body.
Let b denote the body force per unit mass. Consider an elemental volume dv
inside v. The body force of the elemental volume is equal to ρ dv b. Hence,
the total body force of the control volume is
Z
ρ b dv. (6.3)
v

Surface forces are contact forces acting on the boundary surface of the
body. Examples of surface forces are provided by applied forces on the surface
of the body. Surface forces are reckoned per unit area. Let t denote the surface
force per unit area (or surface stress vector). The surface force on an elemental
surface ds of the volume is tds. The total surface force acting on the closed
surface of the volume v is I
t ds. (6.4)
s

Since the stress vector t on the surface is related to the (internal) stress tensor
↔
σ by Cauchy’s formula [see Eq. (4.10)]
↔
t = n̂ · σ , (6.4)

where n̂ denotes the unit normal to the surface, we can express the surface
force as I
↔
n̂ · σ ds.
s

Using the divergence theorem, we can write

I Z
↔ ↔
n̂ · σ ds = ∇ · σ dv. (6.5)
s v

The principle of conservation of linear momentum applied to a given mass

of a medium, instantaneously occupying a volume v bounded by surface s and
acted upon by external surface force t per unit area and body force b per unit
mass, requires
D
Z h Z
↔
i
∇ · σ + ρb dv = ρv dv, (6.6)
v Dt v
where v is the velocity vector. Using the Reynolds transort theorem (5.5), we
arrive at
Dv
Z ´ µ
↔
0= ∇ · σ + ρb − ρ dv, (6.7)
v Dt
32 THEORY OF ELASTICITY

which is the global form of the equation of motion. The local form is given by

↔ Dv
∇ · σ + ρb = ρ (6.8)
Dt
In Cartesian rectangular system, we have

∂σji ∂vi
+ ρbi = ρ (i = 1, 2, 3). (6.9)
∂xj ∂t

In the case of static equilibrium, Eq. (6.8) reduces to

↔ ∂σji
∇ · σ + ρb = 0 or + ρbi = 0 (6.10)
∂xj

Equations of motion (6.9) can be derived by directly applying Newton’s

second law of motion to a volume element. Consider the stresses and body
forces on an infinitesimal parallelepiped element of a material body. Figure 6.1
shows the stresses acting on the various faces of the infinitesimal parallelepiped
with dimensions dx1 , dx2 , and dx3 along coordinate lines (x1 , x2 , x3 ). The sum
of all forces in the x1 -direction is given by

∂σ11 ∂σ21
² ³ ² ³
σ11 + dx1 dx2 dx3 − σ11 dx2 dx3 + σ21 + dx2 dx1 dx3
∂x1 ∂x2
∂σ31
² ³
− σ21 dx1 dx3 + σ31 + dx3 dx1 dx2 − σ31 dx1 dx2 + ρf1 dx1 dx2 dx3
∂x3
∂σ11 ∂σ21 ∂σ31
² ³
= + + − ρb1 dx1 dx2 dx3 .
∂x1 ∂x2 ∂x3

By Newton’s second law of motion, the sum of the forces is equal to the product
of mass and acceleration in the x1 -direction

∂v1
(ρdx1 dx2 dx3 ) ,
∂t
where ρ is the density. Thus, upon dividing throughout by dx1 dx2 dx3 , we
obtain
∂σ11 ∂σ21 ∂σ31 ∂v1
+ + + ρb1 = ρ . (6.11)
∂x1 ∂x2 ∂x3 ∂t
 J. N. REDDY 33

Figure 6.1: Stresses on a parallelepiped element.

Similarly, the application of Newton’s second law in the x2 - and x3 -

directions gives respectively

∂σ12 ∂σ22 ∂σ32 ∂v2

+ + + ρb2 = ρ ,
∂x1 ∂x2 ∂x3 ∂t
∂σ13 ∂σ23 ∂σ33 ∂v3
+ + + ρb3 = ρ ,
∂x1 ∂x2 ∂x3 ∂t

or, in index notation

∂σji ∂vi
+ ρbi = ρ (i = 1, 2, 3), (6.12)
∂xj ∂t

which is the same as that in Eq. (6.11). For static equilibrium, we set the
time derivative terms to zero and obtain the equilibrium equations

∂σji
+ ρbi = 0 (i = 1, 2, 3). (6.13)
∂xj

6.2 Symmetry of Stress Tensor

The principle of conservation of angular momentum states that the time
rate of change of the total moment of momentum for a continuum is equal to
vector sum of the moments of external forces acting on the continuum. In the
34 THEORY OF ELASTICITY

absence of body couples (i.e., volume-dependent couples), the principle leads

to the symmetry of the stress tensor.
A mathematical form of the principle is

D
I Z Z
(x × t)ds + (x × ρb)dv = (x × ρv)dv (6.14)
s v Dt v

In indicial notation (kth component),

D
I Z Z
eijk xi tj ds + ρeijk xi bj dv = ρeijk xi vj dv (6.15)
s v Dt v

We use several steps to simply the expression. First replace tj with tj = np σpj .
Then transform the surface integral to a volume integral and use the Reynold’s
transport theorem for the material time derivative of a volume integral to
obtain

D
Z Z Z
eijk (xi σpj ),p dv + ρeijk xi bj dv = ρeijk (xi vj ) dv
v v v Dt

Carrying out the indicated differentiations and noting Dxi /Dt = vi , we obtain

Dvj
Z Z ² ³
eijk (xi σpj,p + δip σpj + ρxi bj ) dv = ρeijk vi vj + xi dv
v v Dt

Dvj
Z ´ ² ³ µ
eijk xi σpj,p + ρbj − ρ + σij dv = 0
v Dt
or
eijk σij = 0 (6.16)
which implies that σij = σji . That is, the matrix of the stress components is
symmetric:
σ23 = σ32 , σ31 = σ13 , σ12 = σ21 .
Thus, there are only six stress components that are independent.
The symmetry of the stress tensor can also be established using Newton’s
second law for moments. Consider the moment of all forces acting on the
parallelepiped about the x3 −axis (see Figure 6.1). Using the right-handed
screw rule for positive moment, we obtain

∂σ12 dx1 dx1

´² ³ µ
σ12 +
dx1 dx2 dx3 + (σ12 dx2 dx3 )
∂x1 2 2
∂σ21 dx2 dx2
´² ³ µ
− σ21 + dx2 dx1 dx3 − (σ21 dx1 dx3 ) = 0.
∂x2 2 2
 J. N. REDDY 35

1
Dividing throughout by 2 dx1 dx2 dx3 and taking the limit dx1 → 0 and
dx2 → 0, we obtain
σ12 − σ21 = 0.
Similar considerations of moments about the x1 -axis and x2 -axis give,
respectively, the relations

σ23 − σ32 = 0, σ13 − σ31 = 0.

Example 6.1:
Given the following state of stress (σij = σji ),

σ11 = −2x21 , σ12 = −7 + 4x1 x2 + x3 , σ13 = 1 + x1 − 3x2 ,

σ22 = 3x21 − 2x22 + 5x3 , σ23 = 0, σ33 = −5 + x1 + 3x2 + 3x3 ,

determine the body force components for which the stress field describes a state of static
equilibrium.

Solution: The body force components are

∂σ11 ∂σ12 ∂σ13

° ±
ρb1 = − + + = −[(−4x1 ) + (4x1 ) + 0] = 0,
∂x1 ∂x2 ∂x3
∂σ12 ∂σ22 ∂σ23
° ±
ρb2 = − + + = −[(4x2 ) + (−4x2 ) + 0] = 0,
∂x1 ∂x2 ∂x3
∂σ13 ∂σ23 ∂σ33
° ±
ρb3 = − + + = −[1 + 0 + 3] = −4.
∂x1 ∂x2 ∂x3

Thus the body is in static equilibrium for the body force components ρb1 = 0, ρb2 = 0, and
ρb3 = −4.

Example 6.2:
We wish to determine the principal stresses and principal strains associated with the stress
tensor "
2 1 0
#
[σ] = 1 4 1 psi.
0 1 2
The characteristic equation is obtained by setting det (σij − λ δij ) to zero:
¬ ¬
¬2 − λ 1 0 ¬
¬ 1 4−λ 1 ¬ = (2 − λ)[(4 − λ)(2 − λ) − 1] − 1 · (2 − λ) = 0,
¬ ¬
¬ 0 1 2 − λ¬
or
(2 − λ)[(4 − λ)(2 − λ) − 2] = 0.
Hence, the eigenevalues (or principal stresses) are
√ √
λ1 = 3 + 3 = 4.7321 psi, λ2 = 3 − 3 = 1.2679 psi, λ3 = 2 psi.
36 THEORY OF ELASTICITY

The eigenvector (or principal plane) corresponding to λ3 = 2 psi, for example, is

calculated as follows. From (σij − λ3 δij )Aj = 0 we have
"
2−2 1 0
#(
A1
) ( )
0
1 4−2 1 A2 = 0 .
0 1 2−2 A3 0

This gives
A2 = 0, A1 = −A3 .
Using A21 + A22 + A23 = 1, we obtain

Ý 3 = ± √1 (1, 0, −1),
A for λ3 = 2.
2
√
Similarly, the eigenvectors corresponding to λ1,2 = 3 ± 3 are given by
√
√ ¡ ¡ √
Ý 1 = ± (3 − 3) 1, 1 + 3 , 1 ,
   
A for λ1 = 3 + 3,
12
√
(3 + 3)     √ ¡ ¡ √
Ý
A2 = ± 1, 1 − 3 , 1 , for λ2 = 3 − 3.
12

7. Constitutive Equations
7.1 General Introduction
The kinematic relations and the mechanical (e.g., conservation of mass and
momenta) and thermodynamic principles (e.g., conservation of energy) are
applicable to any continuum irrespective of its physical constitution. The
kinematic variables such as the strains and the kinetic variables such as the
stresses were introduced independent of each other. In this section we shall
consider equations characterizing the individual material and its reaction to
applied loads, and these equations relate the kinetic variables to the kinematic
variables. They are called the constitutive equations. The formulation of the
constitutive equations for a given material is guided by certain rules (i.e.,
constitutive axioms). We will not discuss them here but will review the linear
constitutive relations for solids.
Materials for which the constitutive behavior is only a function of the
current state of deformation are known as elastic. In the special case in which
the work done by the stresses during a deformation is dependent only on the
initial state and the current configuration, the material is called hyperelastic.
A material body is said to be homogeneous if the material properties
are the same throughout the body (i.e., independent of position). In a
heterogeneous body, the material properties are a function of position. An
anisotropic body is one that has different values of a material property in
different directions at a point, i.e., material properties are direction-dependent.
 J. N. REDDY 37

An isotropic body is one for which every material property in all directions at a
point is the same. An isotropic or anisotropic material can be nonhomogeneous
or homogeneous.
A material body is said to be ideally elastic when, under isothermal
conditions, the body recovers its original form completely upon removal of
the forces causing deformation, and there is a one-to-one relationship between
the state of stress and the state of strain in the current configuration. The
constitutive equations described here do not include creep at constant stress
and stress relaxation at constant strain. Thus, the material coefficients
that specify the constitutive relationship between the stress and strain
components are assumed to be constant during the deformation. This does
not automatically imply that we neglect temperature effects on deformation.
We account for the thermal expansion of the material, which can produce
strains or stresses as large as those produced by the applied mechanical forces.
Here, we discuss the constitutive equations of linear elasticity (i.e., relations
between stress and strain are linear) for the case of infinitesimal deformation
(i.e., |∇u| << 1). Hence, we will not distinguish between various measures of
stress and strain, and use S ≈ σ for the stress tensor and E ≈ ε for strain tensor
in the material description used in solid mechanics. The linear constitutive
model for infinitesimal deformation is referred to as the generalized Hooke’s
law.
Before we discuss the generalized Hooke’s law, a comment is in order about
the so-called principle of material frame-indifference. Functions and fields
whose values are scalars, vectors, or tensors are called frame-indifferent or
objective if both the dependent and independent vector and tensor variables
transfor according to the following equations:

1. Events, x, t : x∗ = c(t) + Q(t) · x

2. Vectors, v : v∗ = Q(t) · v
3. Second-order tensors, S : S∗ = Q(t) · S · QT
4. Deformation gradient tensor, F : F∗ = Q(t) · F
(F transforms like a vector)

Here quantities without an asterisk refers to a frame of reference F with origin

O and those with an asterisk (*) refere to another frame of reference F ∗ with
origin O∗, c(t) is a constant vector from O to O∗, and Q(t) is the orthogonal
rotation tensor that rotates frame F ∗ into frame F . The frame-indifference
referes to the invariance of quantities with respect to frames of reference. For
example, x and x∗ refer to the same motion, but mathematically x* is the
motion obtained from x by superposition of a rigid transformation (rotation)
and shift (translation). One can show that the velocity and acceleration
vectors are not objective.
38 THEORY OF ELASTICITY

7.2 Generalized Hooke’s Law

Suppose that the reference configuration has a (residual) stress state of
σ 0 . Then if the stress components are assumed to be linear functions of the
components of strain, then the most general form of the linear constitutive
equations for infinitesimal deformations is

σ = C : ε + σ0, 0
σij = Cijk` εk` + σij , (7.1)

where C is the fourth-order tensor of material parameters and is termed

stiffness tensor. There are, in general, 34 = 81 scalar components of a fourth-
order tensor. The number of independent components of C are considerable
less because of the symmetry of σ, symmetry of ε, and symmetry of C, as
discussed next.
In the absence of body moments, the principle of conservation of linear
momentum requires the stress tensor to be symmetric, σij = σji . Then
it follows from Eq. (7.1) that Cijk` must be symmetric in the first two
subscripts. Hence the number of independent material stiffness components
reduce to 6(3)2 = 54. Since the strain tensor is symmetric (by its definition),
εij = εji , then Cijk` must be symmetric in the last two subscripts as well,
further reducing the number of independent material stiffness components to
6 × 6 = 36.
If we also assume that the material is hyperelastic, i.e., there exists a
strain energy density function U0 (εij ) such that

∂U0 0
σij = = Cijk` εk` + σij , (7.2)
∂εij

we have
∂ 2 U0
= Cijk` .
∂εij ∂εk`
Since the order of differentiation is arbitrary, ∂ 2 U0 /∂εij ∂εk` = ∂ 2 U0 /∂εk` ∂εij ,
it follows that Cijk` = Ck`ij . This reduces the number of independent material
stiffness components to 21. To show this we express Eq. (7.1) in an alternate
form using single subscript notation for stresses and strains and two subscript
notation for the material stiffness coefficients:

σ1 = σ11 , σ2 = σ22 , σ3 = σ33 , σ4 = σ23 , σ5 = σ13 , σ6 = σ12 ,

ε1 = ε11 , ε2 = ε22 , ε3 = ε33 , ε4 = 2ε23 , ε5 = 2ε13 , ε6 = 2ε12 . (7.3a)

11 → 1 22 → 2 33 → 3 23 → 4 13 → 5 12 → 6. (7.3b)
It should be cautioned that the single subscript notation used for stresses
and strains and the two-subscript components Cij render them non-tensor
 J. N. REDDY 39

components (i.e., σi , εi , and Cij do not transform like the components of a

vector or tensor) The single subscript notation for stresses and strains is called
the engineering notation or the Voigt-Kelvin notation. Equation (7.1) now
takes the form
σi = Cij εj + σi0 (7.4a)
where summation on repeated subscripts is implied (now from 1 to 6). In
matrix ntation Eq. (7.4a) can be written as

σ C11 C12 C13 C14 C15 C16  ε1   σ10 

      
 1

 σ20 
    
 σ2 




  C21 C22 C23 C24 C25 C26   ε2 
   
       0



σ3   C31 C32 C33 C34 C35 C36  ε3
   
σ3 
= + 0 . (7.4b)

 σ4 
  C41
 C42 C43 C44 C45 C46 
 ε4 
   σ4 
 σ50 
     
 σ5 
  C51 C52 C53 C54 C55 C56  
 ε5 

   
  
      0 
σ6 C61 C62 C63 C64 C65 C66 ε6 σ6

Now the coefficients Cij must be symmetric (Cij = Cji ) by virtue of the
assumption that the material is hyperelastic. Hence we have 6 + 5 + 4 + 3 + 2 +
1 = 21 independent stiffness coefficients for the most general elastic material.
We assume that the stress-strain relations (7.4a,b) are invertible. Thus,
the components of strain are related to the components of stress by

εi = Sij σj + ε0i , ε0i = −Sij σj0 (7.5a)

where Sij are the material compliance parameters with [S] = [C]−1 (the
compliance tensor is the inverse of the stiffness tensor: S = C−1 ). In matrix
form Eq. (7.5a) becomes

ε1  S11 S12 S13 S14 S15 S16  σ1   ε01 

      

ε02 
     


 ε2 

  S21 S22 S23 S24 S25 S26 

 σ2 
 



 


      0
  
ε3   S31 S32 S33 S34 S35 S36  σ 3 ε 3

=  + 0 . (7.5b)

 ε4  S41
 S42 S43 S44 S45 S46 
 σ4  
 ε4 
 ε05 
     


 ε
 5



S
51 S52 S53 S54 S55 S56  
 σ5 



 
   0


ε6 S61 S62 S63 S64 S65 S66 σ6 ε6

In the following discussion we assume that the reference configuration is

stress free, σi0 = 0 and ε0i = 0.

7.3 Material Symmetry

Further reduction in the number of independent stiffness (or compliance)
parameters comes from the so-called material symmetry. Suppose that
(x1 , x2 , x3 ) denote the coordinate system with respect to which Eqs. (7.4a,b)
and (7.5a,b) are defined. We shall call them material coordinate system. The
40 THEORY OF ELASTICITY

coordinate system (x, y, z) used to write the equatins of motion and strain-
displacement equations will be called the problem coordinates to distinguish
them from the material cordinate system. Note that the phrase ‘material
cordinates’ used in connection with the material decription should not be
confused with the present term. In the remaining discussion, we will use the
material desciption for everything, but we may use one material coordinate
system, say (x, y, z), to describe the kinematics as well as stress state in
the body and another material coordinate system (x1 , x2 , x3 ) to describe the
stress-strain behavior. Both are fixed in the body, and the two systems are
oriented with respect to each other. When elastic material parameters at a
point have the same values for every pair of coordinate systems that are mirror
images of each other in a certain plane, that plane is called a material plane
of symmetry (e.g., symmetry of internal structure due to crystallographic
form, regular arrangement of of fibers or molecules, etc.). We note that
the symmetry under discussion is a directional property and not a positional
property. Thus, a material may have certain elastic symmetry at every point
of a material body the properties may vary from point to point. Positional
dependence of material properties is what we called the inhomogeneity of the
material.
In the following we discuss various planes of symmetry and forms of
associated stress-strain relations. Note that use of the tensor components
of stress and strain is necessary because the transformations are valid only
for the tensor components. The fourth-order tensor, for example, transforms
according to the formula
0
Cijkl = `ip `jq `kr `ls Cpqrs (7.6)
where `ij are the direction cosines associated with the coordinate systems
(x1 , x2 , x3 ) and (x01 , x20 , x03 ), and Cijkl
0 and Cpqrs are the components of the
fourth-order tensor C in the primed and unprimed coordinates systems,
respectively.

7.4 Monoclinic Materials

When the elastic coefficients at a point have the same value for every pair
of coordinate systems which are the mirror images of each other with respect
to a plane, the material is called a monoclinic material. For example, let
(x1 , x2 , x3 ) and (x01 , x02 , x03 ) be two coordinates systems, with the x1 , x2 -plane
plane parallel to the plane of symmetry. Choose x03 -axis such that x03 = −x3
(never mind about the left-handed coordinate system as it does not affect the
discussion) so that one system is the mirror image of the other. The definitions
and sign conventions of the stress and strain components show that
0 0
σ23 = −σ23 , σ31 = −σ31 , ε023 = −ε23 , ε031 = −ε31
 J. N. REDDY 41

or, in single-subscript notation

σ40 = −σ4 , σ50 = −σ5 , ε04 = −ε4 , ε05 = −ε5

while all ther independent stress and strain components remain unchanged
in value by the change from one coordinate system to the other. Using the
stress-strain relations of the form in (7.4b), we can write

σ10 = C11 ε01 + C12 ε20 + C13 ε03 + C14 ε40 + C15 ε05 + C16 ε06
σ1 = C11 ε1 + C12 ε2 + C13 ε3 − C14 ε4 − C15 ε5 + C16 ε6

But we also have

σ1 = C11 ε1 + C12 ε2 + C13 ε3 + C14 ε4 + C15 ε5 + C16 ε6

Note that the elastic parameters Cij are the same for the two coordinate
systems because they are the mirror images in the plane of symmetry. From
the above two equations (subtract ne from the other) we arrive at

C14 ε4 + C15 ε5 = 0 for all values of ε4 and ε5

The above equation holds only if C14 = 0 and C15 = 0. Similar discussion
with the two alternative expressions of the remaining stress components yield
C24 = 0 and C25 = 0; C34 = 0 and C35 = 0; and C46 = 0 and C56 = 0.
Thus out of 21 material parameters, we only have 21 − 8 = 13 independent
parameters, as indicated below

C11 C12 C13 0 0 C16

 
 C12 C22 C23 0 0 C26 
 
 C13 C23 C33 0 0 C36 
[C] =   (7.8)
 0
 0 0 C44 C45 0 
 0 0 0 C45 C55 0 
C16 C26 C36 0 0 C66

Note that monoclinic materials exhibit shear-extensional coupling, i.e., a shear

strain can produce a normal stress; for example, σ11 = C16 ε6 = 2C16 ε12 .
Therefore, the principal axes of stress do not coincide with those of strain.
The result in Eq. (7.8) can also be obtained also using the following
transformation matrix (which converts the unprimed coordinate system to
the primed one) in Eq. (7.6):
 
1 0 0
[L] = 0 1
 0  (or `11 = `22 = 1, `33 = −1, `ij = 0 for i 6= j) (7.9)
0 0 −1
42 THEORY OF ELASTICITY

7.4 Orthotropic Materials

When three mutually orthogonal planes of material symmetry exist, the
number of elastic coefficients is reduced to 9 using arguments similar to
those given for single material symmetry plane, and such materials are called
orthotropic. The stress-strain relations for an orthotropic material take the
form
 σ1  C11 C12 C13 0 0 0  ε1 
    
   



 σ2 



 C12 C22 C23
 0 0 0 


 ε2 

 

σ3
  
C13 C 23 C 33 0 0 0  ε 3

=  . (7.10)

 σ4 

 0
 0 0 C44 0 0  ε4 
   
 σ5  C55  ε5 


   0
 0 0 0 0  



 
σ6 0 0 0 0 0 C66 ε6
The transfrmation matrices associated with the planes of symmetry are
−1
     
1 0 0 0 0 1 0 0
[L(1) ] =  0 1 0  ; [L(2) ] =  0 0  ; [L(3) ] =  0 −1 0 
1
0 0 −1 0 1 0 0 0 1
(7.11)
Most simple characterization tests are performed with a known load or
stress. Hence, it is convenient to write the inverse of relations in (7.10). The
strain-stress relations of an orthotrpic material are given by
ε S11 S12 S13 0 0 0  σ1 
    
 1
   
 ε2  S12 S22 S23




  0 0 0   σ2 
 

   

ε3   S13 S23 S33 0 0 0  σ3 
 
= (7.11)
ε4  0 0 0 S44 0 0  σ 
 4
 

 
   
 ε5  0 0 0 0 S55 0   σ5 

 
   

   
ε6 0 0 0 0 0 S66 σ6
where
2
S22 S33 − S23 S13 S23 − S12 S33
C11 = C12 =
S S
2
S33 S11 − S13 S12 S23 − S13 S22
C22 = C13 =
S S
2
S11 S22 − S12 S12 S13 − S23 S11
C33 = C23 =
S S
1 1 1
C44 = C55 = C66 =
S44 S55 S66
2 2 2
S = S11 S22 S33 − S11 S23 − S22 S13 − S33 S12 + 2S12 S23 S13 (7.12)

Most often, the material properties are determined in a laboratory in

terms of the engineering constants such as Young’s modulus, shear modulus,
 J. N. REDDY 43

and so on. These constants are measured using simple tests like uniaxial
tension test or pure shear test. Because of their direct and obvious physical
meaning, engineering constants are used in place of the more abstract stiffness
coefficients Cij and compliance coefficients Sij . Next we discuss how to relate
Sij to the engineering constants.
One of the consequences of linearity (both kinematic and material
linearizations) is that the principle of superposition applies. That is, if the
applied loads and geometric constraints are independent of deformation, the
sum of the displacements (and hence strains) produced by two sets of loads
is equal to the displacements (and strains) produced by the sum of the two
sets of loads. In particular, the strains of the same type produced by the
application of individual stress components can be superposed. For example,
(1)
the extensional strain ε11 in the material coordinate direction x1 due to the
stress σ11 in the same direction is σ11 /E1 , where E1 denotes Young’s modulus
(2)
of the material in the x1 direction. The extensional strain ε11 due to the stress
σ22 applied in the x2 direction is −ν21 σ22 /E2 , where ν21 is the Poisson ratio

ε11
ν21 = −
ε22

and E2 is Young’s modulus of the material in the x2 direction. Similarly, σ33

(3)
produces a strain ε11 equal to −ν31 σ33 /E3 . Hence, the total strain ε11 due to
the simultaneous application of all three normal stress components is

(1) (2) (3) σ11 σ22 ν21 σ33 ν31

ε11 = ε11 + ε11 + ε11 = − − (a)
E1 E2 E3

where the direction of loading is denoted by the superscript. Similarly, we can

write
σ11 ν12 σ22 σ33 ν32
ε22 = − + − (b)
E1 E2 E3
σ11 ν13 σ22 ν23 σ33
ε33 = − − + (c)
E1 E2 E3
The simple shear tests with an orthotropic material give the results

σ12 σ13 σ23

2ε12 = , 2ε13 = , 2ε23 = (d)
G12 G13 G23

6 j) is the change in the right angle between two lines

Recall that 2εij (i =
parallel to the x1 and x2 directions at a point, σij (i = 6 j) denotes the
corresponding shear stress in the xi -xj plane, and Gij (i 6= j) are the shear
44 THEORY OF ELASTICITY

moduli in the xi -xj plane. Writing Eqs. (a)-(d) in matrix form, we obtain
 1
E1 − νE212 − νE313 0 0 0 
ε1  σ 
   

   − νE121 1
− νE323 0 0 0  1

 ε2 
  E2  σ2 

− νE131 − νE232
   
1
0 0 0

    
ε3   E3
 σ3 

=   σ4  , (7.13)

ε4  1



 
  0 0 0 G23 0 0  
 ε   1 σ  

 5
  

 0 0 0 0 G13 0  
 5 

ε6 0 0 0 0 0 1 σ6
G12

where E1 , E2 , E3 are Young’s moduli in 1, 2, and 3 material directions,

respectively, νij is Poisson’s ratio, defined as the ratio of transverse strain
in the jth direction to the axial strain in the ith direction when stressed in the
i-direction, and G23 , G13 , G12 = shear moduli in the 2-3, 1-3, and 1-2 planes,
respectively. Since the compliance matrix [S] is the inverse of the stiffness
matrix [C] and the inverse of a symmetric matrix is symmetric, it follows that
the compliance matrix [S] is also a symmetric matrix. This in turn implies
that the following reciprocal relations hold [see Eq. (7.12)]:
ν21 ν12 ν31 ν13 ν32 ν23
= ; = ; =
E2 E1 E3 E1 E3 E2

or, in short
νij νji
= (no sum on i, j) (7.14)
Ei Ej
for i, j = 1, 2, 3. The 9 independent material coefficients for an orthotropic
material are
E1 , E2 , E3 , G23 , G13 , G12 , ν12 , ν13 , ν23 . (7.15)
It is important to note the difference, for example, between νij and νji for
i 6= j for an orthotropic material. For example the difference between ν12 and
ν21 for an orthotropic material is illustrated in Figure 7.1 with two cases of
uniaxial stress for a square element of length a. First a stress σ is applied in
the x1 -direction as shown in Figure 7.1a. The resulting strains are
(1) σ (1) ν12
ε11 = ε22 = − σ (7.16)
E1 E1

where the direction of loading is denoted by the superscript and negative sign
indicates compression. Next, the same value of stress is applied in the x2 -
direction as shown in Figure 7.1b. The strains are
(2) ν21 (2) σ
ε11 = − σ ε22 = (7.17)
E2 E2
 J. N. REDDY 45

Figure 7.1: Distinction between ν12 and ν21 .

(1) (2)
While it is obvious that ε11 < ε22 if E1 > E2 , we have no clue about the
(1) (2)
relative magnitudes of ε22 and ε11 . However, the displacements associated
with the two loadings are
(1) σ (1) ν12
u1 = a u2 = −a σ (7.18a)
E1 E1
(2) ν21 (2) σ
u1 = −a σ u2 = a σ (7.18b)
E2 E2
(1) (2)
and the reciprocal relation (7.14) gives u2 = u1 , which is the statement of
Betti’s reciprocity theorem (we shall study it in the later chapters).
Comparing Eqs. (7.11) and (7.13), we note that
1 ν12 ν13
S11 = , S12 = − , S13 = −
E1 E1 E1
1 ν23 1
S22 = , S23 = − , S33 =
E2 E2 E3
1 1 1
S44 = , S55 = S66 = (7.19)
G23 G13 G12
46 THEORY OF ELASTICITY

and the stiffness coefficients are given by

1 − ν23 ν23 ν21 + ν31 ν23 ν12 + ν32 ν13

C11 = , C12 = =
E2 E3 ∆ E2 E3 ∆ E1 E3 ∆
ν31 + ν21 ν32 ν13 + ν12 ν23
C13 = =
E2 E3 ∆ E1 E2 ∆
1 − ν13 ν31 ν32 + ν12 ν31 ν23 + ν21 ν13
C22 = , C23 = =
E1 E3 ∆ E1 E3 ∆ E1 E3 ∆
1 − ν12 ν21
C33 = , C44 = G23 C55 = G31 C66 = G12
E1 E2 ∆
1 − ν12 ν21 − ν23 ν32 − ν31 ν13 − 2ν21 ν32 ν13
∆= (7.20)
E1 E2 E3

7.5 Isotropic Materials

When there exist no preferred directions in the material (i.e., the material has
infinite number of planes of material symmetry), the number of independent
elastic coefficients reduces to 2. Such materials are called isotropic. For
isotropic materials we have

E1 = E2 = E3 = E, G12 = G13 = G23 ≡ G, ν12 = ν23 = ν13 ≡ ν. (7.21)

Consequently, Eqs. (7.10) and (7.13), in view of the relations (7.19), (7.20)
and (7.21) [also note that 1 − 3ν 2 − 2ν 3 = (1 + ν)2 (1 − 2ν)], take the form

σ 1−ν ν ν 0 0 0 ε1 
    
 1
 
1−ν
  
 σ2 




  ν ν 0 0 0  
 ε2 

   


σ3   ν ν 1−ν 0 0 0  ε3 
 
= Λ ,

 σ4 

 0
 0 0 1 − 2ν 0 0   ε4 

1 − 2ν
   
 σ5   ε5 

 
  0 0 0 0 0   

   
σ6 0 0 0 0 0 1 − 2ν ε6
(7.22)
ε1  1 −ν −ν 0 0 0 σ1 
    
 
−ν −ν
   



 ε2 

  1 0 0 0  

 σ 2


  
1 
 
ε3 
−ν −ν 1 0 0 0  
σ 3

=   , (7.23)

 ε4 
 E 0 0 0 1+ν 0 0  σ4 
   


 ε
 5



 0 0 0 0 1+ν 0   σ5 





ε6 0 0 0 0 0 1+ν σ6
where
E
Λ= (7.24)
(1 + ν)(1 − 2ν)
 J. N. REDDY 47

Alternatively, the stress-strain relations can be written in more compact

form using the fact that a fourth-order isotropic tensor can be expressed as

Cijk` = λδij δk` + µ (δik δj` + δi` δjk ) (7.25)

where λ and µ are called Lamé constants. Therefore, the stress-strain relation
for the isotropic case takes the form

σij = Cijk` εk` = 2µεij + λεkk δij , σ = 2µε + λ tr(ε)I (7.26)

The strain-stress relations are

1 λ 1 λ
´ µ ´ µ
εij = σij − σkk δij , ε = σ− tr(σ)I (7.27)
2µ (2µ + 3λ) 2µ (2µ + 3λ)

We note the following relations between the Lamé constants λ and µ and
engineering constants E, ν and G for an isotropic material:
µ(3λ + 2µ) λ 2
E= , ν= , K = λ + µ, G=µ (7.28)
λ+µ 2(µ + λ) 3
The following definitions and constitutive relations are of interest in the sequel:
1
mean stress, σ̃ ≡ σii , dilatation, e ≡ εii (7.29)
3
1
deviatoric stress, σ 0 = σ − σ̃I ,deviatoric strain, ε0 = ε − tr(ε) (7.30)
3
2
σii = (3λ + 2µ)εii , σ̃ = Ke, K = λ + µ (7.31)
3
where K is the bulk modulus and µ = G is the shear modulus.
In view of the relations between the Lamé constants and engineering
constants, Eqs. (7.26) and (7.27) can be written in terms of engineering
constants:
E νE E νE
σij = εij + εkk δij , σ = ε+ tr(ε)I
1+ν (1 + ν)(1 − 2ν) 1+ν (1 + ν)(1 − 2ν)
(7.32)
1 1
εij = [(1 + ν)σij − νσkk δij ] , ε = [(1 + ν)σ − ν tr(σ)I] (7.33)
E E
The strain energy density for a linear isotropic material is given by
1 1
U0 = Cijk` εij εk` = σij εij
2 2
1
= (σ11 ε11 + σ22 ε22 + σ33 ε33 + 2σ12 ε12 + 2σ13 ε13 + 2σ23 ε23 ) (7.34)
2
48 THEORY OF ELASTICITY

7.6 Plane-Stress Constitutive Equations

A plane stress state is defined to be one in which all transverse stresses are
negligible. The strain-stress relations of an orthotropic material in plane stress
state can be written as
1
− νE212
 
 
 ε1 

S11 S12 0  σ1 

E1
 0  
 σ1 
ε2 =  S12 S22 0  σ2 =  − νE121 1
0 
 σ2 . (7.35)

E2
ε6 0 0 S66 σ6 1 σ6
     
0 0 G12

The strain-stress relations (7.35) can be inverted to obtain the stress-strain

relations     
 σ1  Q11 Q12 0  ε1 
σ =  Q12 Q22 0  ε2 , (7.36)
 2
σ6 0 0 Q66 ε6
 

where the Qij , called the plane stress-reduced stiffnesses, are given by

S22 E1
Q11 = 2 = ,
S11 S22 − S12 1 − ν12 ν21
S12 ν12 E2
Q12 = 2 = ,
S11 S22 − S12 1 − ν12 ν21
S11 E2
Q22 = 2 = ,
S11 S22 − S12 1 − ν12 ν21
1
Q66 = = G12 . (7.37)
S66

Note that the reduced stiffnesses involve 4 independent material constants,

E1 , E2 , ν12 , and G12 .
The transverse shear stresses are related to the transverse shear strains in
an orthotropic material by the relations
º » ´ µº »
σ4 Q44 0 ε4
= , (7.38)
σ5 0 Q55 ε5

where Q44 = C44 = G23 and Q55 = C55 = G13 . For an isotropic material, Eqs.
(7.36) and (7.38) take the form
  
σ1  1 ν 0 0 0  ε1 
 


σ   ν 1 0 0 0
ε  
 2
  E   2 
0 1−ν
σ4 = 0 2 0 0
 ε4 ,

(7.39)
− 2 

1 ν 1−ν
σ 0 0 0 0 ε5 

 
   
 5 2

   
1−ν
  
σ6 0 0 0 0 2
ε6
 J. N. REDDY 49

A qualitative understanding of the anisotropic behavior of a material

can be obtained by simple tension and shear tests. Application of a normal
stress to a rectangular block of isotropic or orthotropic material leads to only
extension in the direction of the applied stress and contraction perpendicular
to it, whereas an anisotropic material experiences extension in the direction of
the applied normal stress, contraction perpendicular to it, as well as shearing
strain (see Figure 7.2). Conversely, the application of a shearing stress to an
anisotropic material causes shearing strain as well as normal strains. Normal
stress applied to an orthotropic material at an angle to its principal material
directions causes it to behave like an anisotropic material.

Figure 7.2: Deformation of othrotropic and anisotropic rectangular block

under uniaxial tension.

7.7 Thermoelastic Constitutive Relations

When temperature changes occur in the elastic body, we account for the
thermal expansion of the material, even though the variation of elastic
constants with temperature is neglected. When the strains, geometric changes,
and temperature variations are sufficiently small, all governing equations are
linear and superposition of mechanical and thermal effects is possible. The
linear thermoelastic constitutive equations have the form
σj = Cji [−αi (T − T0 ) + εi ], (7.40)
εj = Sji σi + αi (T − T0 ), (7.41)
50 THEORY OF ELASTICITY

where αi (i = 1, 2, 3) are the linear coefficients of thermal expansion, T denotes

temperature, and T0 is the reference temperature of the undeformed body. In
writing Eqs. (7.39) and (7.40), it is assumed that αi and Cij are independent of
strains and temperature. For an isotropic material we have α1 = α2 = α3 ≡ α.
The plane stress constitutive relations for a thermoelastic case are given by

σ1  Q11 Q12 0 0 0  ε1 − α1 ∆T
    
 
 ε2 − α2 ∆T

σ   
 2
 
  Q12 Q22 0 0 0  






σ4 = 
 0 0 Q44 0 0  ε4 , (7.42)
σ  0 0 0 Q55 0  ε5

 
 

 5

  
 

  
σ6 0 0 0 0 Q66 ε6

where ∆T = T − T0 is the temperature change from the reference temperature

T0 and αi (i = 1, 2) are the coefficients of thermal expansion of an orthotropic
material in the xi −coordinate direction.

Example 7.1:
For an isotropic material with E = 71.0 GPa and G = 26.6 GPa, determine the strain tensor
and the strain energy density at a point in a body if the stress tensor at that point is given
Ý} by the following matrix of scalar components:
with respect to the vector basis {e
"
20 −4 5
#
[σ] = −4 0 10 MPa
5 10 15

Solution: First compute Poisson’s ratio using E and G

E 71
ν= −1= − 1 = 0.335
2G 26.6
The strains are
1 1
ε11 = [σ − ν (σ22 + σ33 )] = [20 − 0.335 (0 + 15)] 106 = 0.211 × 10−3 m/m
E 11 71 × 109
1 1
ε22 = [σ − ν (σ33 + σ11 )] = [0 − 0.335 (15 + 20)] 106 = −0.165 × 10−3 m/m
E 22 71 × 109
1 1
ε33 = [σ − ν (σ11 + σ22 )] = [15 − 0.335 (20 + 0)] 106 = 0.117 × 10−3 m/m
E 33 71 × 109
1+ν 1 + 0.335
ε12 = σ12 = (−4)106 = −0.075 × 10−3 m/m
E 71 × 109
1+ν 1 + 0.335
ε13 = σ13 = (5)106 = 0.094 × 10−3 m/m
E 71 × 109
1+ν 1 + 0.335
ε23 = σ23 = (10)106 = 0.188 × 10−3 m/m
E 71 × 109
The strain tensor components in matrix form are
"
211 −75 94
#
[ε] = −75 −165 188 10−6 m/m
94 188 117
 J. N. REDDY 51

The strain energy density for the isotropic material is given by

1
U0 = Cijk` εij εk`
2
1
= [λδij δk` + µ (δik δj` + δi` δjk )] εij εk`
2
1 1
= (λδij εkk + 2µεij ) εij = σij εij
2 2
1
= (σ11 ε11 + σ22 ε22 + σ33 ε33 + 2σ12 ε12 + 2σ13 ε13 + 2σ23 ε23 )
2
1
= [(20)(211) + (0)(−165) + (15)(117) + 2(−4)(−75) + 2(5)(94) + 2(10)(188)] 103
2
= 5637.5 Pa