3.9.

Exercise
P3.1 Derive the expression of the Eulerian strain in Eq. (3.17).

Solution:
From the definistion of the Eulerian strain, it is necessary to define the inverse of the
deformation gradient in terms of the displacement gradient. From the relation X = x -
u,

¶X ¶(x - u)
F-1 = = = 1 - n u
¶x ¶x
Then, the Eulerian strain in Eq. (4.17) can be written as

e = 21 (1 - F-T F-1 )
= 1
2 (1 - (1 -  u) (1 -  u))
n
T
n
T T
= (n u + n u - n u n u)
1
2

P3.2 Derive the relation in volume change in Eq. (3.26) for an infinitesimal hexahedron
whose edges are initially parallel to the coordinate directions.

Solution:
Let E1, E2, and E3 be the unit basis vectors for the coordinate system, and the legths of the
hexahedron are dX1, dX2, and dX3. Then, the three edges at the undeformed state can be
written as

d X1 = d X 1 E 1 , dX2 = d X 2 E 2 , dX2 = d X 2 E 2

With these three vectors, the volume of the infinitesimal hexahedron can be calculated by

dV0 = dX1 ⋅ (dX2 ´ dX3 ) = dX 1dX 2 dX 3 E1 ⋅ (E2 ´ E 3 ) = dX1dX 2 dX 3

Now, after deformation, the three edges are deformed to

¶x ¶x ¶x
dx1 = FdX1 = dX 1 , dx2 = FdX2 = dX 2 , dx 3 = FdX 3 = dX 3
¶X 1 ¶X 2 ¶X 3

The deformed volume becomes

¶x æ ¶x ¶x ö÷÷
dVx = dx1 ⋅ (dx2 ´ dx3 ) = dX1dX 2dX 3 ⋅ ççç ´ ÷
¶X1 çè ¶X 2 ¶X 3 ø÷

The determinant of the deformation gradient can be written as

 ¶x æç ¶x ¶x ö÷÷
J = det(F) = ⋅ çç ´ ÷
¶X1 çè ¶X 2 ¶X 3 ÷ø

Thus, the deformed volume becomes

dVx = J dV0

P3.3 Consider a square block under oscillating simple shear deformation. The relation
between undeformed and deformed geometry is given as

x 1 = X 1 + aX 2 sin wt, x 2 = X 2, x3 = X3

Calculate the deformation gradient and the change in volume.

Solution:
The deformation gradient can be calculated from its definition as
é1 a sin wt 0ù
ê ú
F = êê0 1 0úú
ê ú
êë0 0 1ú
û
The determinant of the deformation gradient is J = det(F) = 1. Thus, the volume of the
block does not change due to the given deformation. ▄

P3.4 Many materials often show very different behavior between volume-changing
deformation and volume-preserving deformation. The former is called dilatation, while
the latter is called distortion. In such a case, it is necessary to separate the dilatational and
distortional parts from the deformation gradient. Let F be the deformation gradient,
decompose it such that F = Fv⋅Fd, where Fv is the dilatational part and Fd is the
distortional part. Calculate Fv and Fd using the third invariant of the deformation
gradient.

Solution:
The determinant of the product of two tensors satisfies the following relation:

J = det(F) = det(Fv )det(Fd )

From the definition, det(Fv ) = J and det(Fd ) = 1 . In the view of Example 3.1, the
dilatational deformation gradient must have the following form:
 Index 3
él 0 0 ù
ê ú
Fv = êê 0 l 0 úú
ê ú
êë 0 0 l úû

where the principal stretch can be written in terms of J by

1
det(Fv ) = l 3 = J  l =J3

Thus, the following Fv is obtained:

1
Fv = J 3 1

and the distortional part of the deformation gradient can be written as

-1
Fd = Fv-1 ⋅ F = J 3 F

P3.5 Repeat Problem P3.4 for the Cauchy-Green deformation tensor; i.e., decompose C
= Cv⋅Cd.

Solution:
From the definition of the Cauchy-Green tensor C = FTF, the determinant of C becomes
det(C) = det(FT)det(F) = J2. The volumetric part of the Cauchy-Green tensor comes
from the volumetric part of the deformation gradient: Cv = FvT Fv , whose determinant
becomes
2
det(Cv ) = det(FvT )det(Fv ) = J 3

Thus, the following Cv is obtained:

2
Cv = J 3 1

and the distortional part of the Cauchy-Green tensor can be written as

-2
Cd = Cv-1 ⋅ C = J 3 C

▄
P3.6 Consider a bar with a square cross section in the figure under uniaxial tension
loading. The principal stretch in X1 direction is given by l > 1. When material is
incompressible, compare X1 component of normal strain using Lagrangian, Eulerian, and
engineering strains.

X2

F
X1
X3

Solution:
Since the bar will maintain rectangular shape, there is no shear deformation. In addition,
since both X2 and X3 directions are unconstrained, and the cross-sectional geometries are
identical, the principal stretches in these two directions will be the same. Thus, the
relation between undeformed and deformed geometries can be written as

x 1 = l X 1, x 2 = l2X 2 , x 3 = l2X 3

Since the material is incompressible, the volume should be preserved:

l1l22 = 1  l2 = l-1/2

Thus, the deformation gradient can be obtained as

él 0 0 ùú
ê
F = êê 0 l-1/2 0 úú
ê ú
ê0 0 l-1/2 ú
ë û
And the right Cauchy-Green deformation tensor can be obtained as
él 2 0 0 ùú
ê
C = êê 0 l-1 0 úú
ê ú
ê0 0 l-1 ú
ë û
The Lagrangian strain becomes
él 2 - 1 0 0 ùú
ê
1ê
E= ê 0 l-1 - 1 0 úú
2ê ú
ê 0 0 l-1 - 1ú
ë û
For the Eulerian strain, the inverse of the left Cauchy-Green deformation tensor can be
calculated by
 Index 5
él-2 0 0 ù
ê ú
G-1 = êê 0 l 0 úú
ê ú
êë 0 0 l úû

And the Eulerian strain becomes

é1 - l-2 0 0 ùú
ê
1
e = êê 0 1-l 0 úú
2ê ú
êë 0 0 1 - lú
û
By differentiating the following displacement

u1 = (l - 1)X 1, u2 = (l2 - 1)X 2 , u 3 = (l2 - 1)X 3

the engineering strain can be obtained as

él - 1 0 0 ù
ê ú
e = êê 0 l-1/2 - 1 0 ú
ú
ê -1/2 ú
ê 0 0 l - 1ú
ë û
Below are the normal strains in the X1 direction from the three different strains:

1 1
E11 = (l 2 - 1) e11 = (1 - l-2 ) e11 = l - 1
2 2
The figure below shows the difference between these three strain components. When the
strain is small; i.e., l » 1, all three strains are similar. However, the difference becomes
large as the strain increase.
 ▄

P3.7 A four node square element undergoes large displacement and rotation in the XY
plane, as shown in the figure. The node initially at the origin is moved to (1, 1 – sinp/4)
and the element is rotated by 45°. Calculate the deformation gradient. Compute the
Lagrangian strain and demonstrate that no strain occurs during rigid body motion.
y

Current
1

Initial

x
1

Solution:
From geometry the coordinates in the initial (X,Y) and the deformed (x,y) configurations
are as follows.

Node X Y x y
1 0 0 1 1 – sinp/4
2 1 0 1 + sinp/4 1
3 1 1 1 1 + sinp/4
4 0 1 1 – sinp/4 1

We first need to develop mapping of the current configuration in terms of the initial
configuration. A systematic way to develop this mapping in the finite element context is
 Index 7
to use the interpolation functions to map the given configurations into a 2×2 reference
element in Chapter 1. The interpolation functions are as follows.
ìï N = 14 (1 - s )(1 - t )
ïï 1
ïïN = 14 (1 + s )(1 - t )
ïí 2
ïïN 3 = 14 (1 + s )(1 + t )
ïï
ïïîN 4 = 14 (1 - s )(1 + t )

Using these interpolation functions, the initial configuration is mapped as follows.

s +1
X = N 1x 1 + N 2x 2 + N 3x 3 + N 4x 4 =
2
t +1
Y = N 1y1 + N 2y2 + N 3y 3 + N 4y 4 =
2
By inverting this mapping, the s and t can be written in terms of X and Y as follows.
s = 2X - 1
t = 2Y - 1

The current configuration can also be mapped into s, t using the interpolation functions.

x=
1
4
( 2s - 2t + 4 )
y=
1
4
( 2s + 2t + 4)

Now the desired mapping between the initial and the current configurations can be
written by substituting for s, t in terms of X, Y giving

x=
1
4
( 2(2X - 1) - 2(2Y - 1) + 4 )
y=
1
4
( 2(2X - 1) + 2(2Y - 1) + 4)
z =Z
The deformation gradient can now be easily computed by direct differentiation.
é1 - 1
0ùú
ê 2 2
F = êê 12 1
2
0úú
ê ú
êë 0 0 1ú
û
Using the deformation gradient, the matrix of Lagrangian strains is as follows.
 é 0 0 0ù
ê ú
1 T
E = (F F - 1) = ê 0 0 0úú
ê
2 ê ú
êë 0 0 0úû

This matrix mathematically states the obvious fact that in this example any differential
line segment in the original configuration has the same length in the current
configuration. This example demonstrates that Lagrangian strains are invariant with
respect to rigid body rotations and displacements. ▄

P3.8 A square plane strain element is deformed as shown in the figure. The relation
between deformed and undeformed coordinates is given as

x 1 = X 1 - aX 1X 2 , x 2 = X2, x3 = X3

Compare the engineering strain and Lagrangian strain. Show that the two strain measures
become identical when ‘a’ approach zero.
2
a a

X2

2 X1

a a

Original element
Deformed element

Solution:
The displacement vector can be written as

u = [-aX1X2, 0, 0]T

For linear elastic model, the engineering strain tensor can be defined as
é -aX - 21 aX1 ùú
e = sym(0 u) = êê 1 2
êë - 2 aX1 0 úú
û
Thus, the strain varies linearly.
For the geometric nonlinear model, the deformation gradient and Lagrangian strain
can be calculated by
é 1 - aX2 -aX1 ù
F = 1 + 0u = êê ú
ú
êë 0 1 úû
 Index 9

é -aX (1 - 1 aX ) - 1 aX (1 - aX ) ù é 1 a 2X 2 1 a 2X X ù
E= 1 (FT F ê
- 1) = ê 1 2 2 2 2
1
1
2 2
2 ú
ú = e + êê 1 2 2 2 2 1 2
1 a 2X 2
ú
ú
êë - 2 aX1(1 - aX 2 )
2 a X1
2 úû êë 2 a X1X 2 2 1 úû

Note that the the Lagrangian strain is the engineering strain plus nonlinear terms, which
will approach zero fast when ‘a’ approaches zero. ▄

P3.9 The relation between deformed and undeformed coordinates for pure bending of a
plane strain solid is given as

x 1 = X 1 - aX 1X 2 , x 2 = X 2 + 21 aX 12 , x3 = X3

Compare the engineering strain and Lagrangian strain. Show that the two strain measures
become identical when ‘a’ approach zero.

Solution:
The displacement vector can be written as

u = [-aX1X2, 21 aX12, 0]T

For linear elastic model, the engineering strain tensor can be defined as
é -aX2 0 ù
e = sym(0u) = êê ú
ú
0 0
ëê úû

Thus, e11 is only non-zero component and varies linearly with respect to X 2 .
For the geometric nonlinear model, the deformation gradient and Lagrangian strain
can be calculated by
é 1 - aX2 -aX1 ù
F = 1 + 0u = êê ú
ú
êë aX 1 1 úû

E = 21 (FT F - 1)
é -aX + 1 a 2 (X 2 + X 2 ) 1 a 2X X ù
= êê 2 2
1 a 2X X
1 2 2 1 2ú
1 a 2X 2 ú
êë 2 1 2 2 1 úû
éX2 + X2 X X ù
= e + 21 a 2 êê 1 2 1 2ú
X1X 2 X12 úú
ëê û
Note that the Lagrangian strain is the engineering strain plus nonlinear terms, which will
approach zero fast when ‘a’ approaches zero. ▄

P3.10 In the small deformation theory, the volumetric strain (dVx - dV0)/dV0 is
approximated by e11 + e22 + e33, while in the large deformation theory, it is given by J -
1. Show that when the deformation is small, the latter can be approximated by the former.

Solution:
The deformation gradient can be written in terms of displacement gradient as
é ù
ê1 + ¶u1 ¶u1 ¶u1 ú
ê ¶X1 ¶X 2 ¶X 3 úú
ê
ê ¶u ¶u2 ¶u2 úú
F = êê 2
1+
ê ¶X 1
¶X 2 ¶X 3 úú
ê ¶u ¶u 3 ¶u 3 úú
ê 3
1+
ê ¶X ¶X 2 ¶X 3 úûú
ëê 1

The determinant of the deformation gradient can be written as

æ ¶u1 ö÷ éêæç ¶u2 öæ÷÷ çç ¶u 3 ÷ö ¶u2 ¶u 3 ùú

det F = ççç1 + ÷÷ ç1 + ÷ 1 + ÷÷ -
çè ¶X1 ÷ø êêççè ¶X 2 ÷øèçç ¶X 3 ÷ø ¶X 3 ¶X 2 úú
ë û
¶u1 éê ¶u2 ¶u 3 ¶u2 æç ¶u 3 ö÷ùú
+ - ç1 + ÷÷
¶X 2 êê ¶X 3 ¶X1 ¶X1 çèç ¶X 3 ÷øúú
ë û
é
¶u ¶u ¶u 3 ¶u 3 ç æ ¶u2 ÷ú ö ù
+ 1 êê 2 - çç1 + ÷÷
¶X 3 êë ¶X1 ¶X 2 ¶X1 çè ¶X 2 ÷øúú
û
¶u1 ¶u2 ¶u 3
= 1+ + + + H.O.T.
¶X1 ¶X 2 ¶X 3

When the deformation is small, the higher-order terms will approach zero quickly. Thus
the volumetric strain can be approximated by

¶u1 ¶u 2 ¶u 3
J -1 = + + » e11 + e22 + e33
¶X 1 ¶X 2 ¶X 3

P3.11 An initially straight beam AB is bent into a circular arc A'B' as shown in the
figure. The deformation is specified as

p(1 - X1 ) p(1 - X1 )
x 1 = g(X 2 )cos , x 2 = g(X 2 )sin , x 3 = X3
2 2
 Index 11
where g(X2) is a simple function of X2. (a) Find the deformation gradient in terms of
g(X2). (b) If the volume of the beam does not change, find g(X2). (c) Using g(X2) in (b),
find U, Q, and V.
X2
Undeformed beam

1
Deformed beam

X1
1

Solution:
(a) For the given deformation, the deformation gradient can be obtained as
é p ù
ê g(X )sin p (1 - X ) dg p
cos (1 - X1 ) 0ú
ê 2 2
2 1
dX 2 2 ú
ê ú
ê p p dg p ú
F = ê- g(X 2 )cos (1 - X1 ) sin (1 - X1 ) 0ú
ê 2 2 dX 2 2 ú
ê ú
ê 0 0 1ú
ê ú
êë úû

(b) In order to preserve the volume, the determinant of the deformation gradient must be
one.

p dg
det F = g(X 2 ) =1
2 dX 2

Using the separation of variables,

2
gdg = dX 2
p
Integrating both sides

1 2 2
g = X2
2 p
Thus, the expression of g(X2) is obtained as
 4X 2
g(X 2 ) =
p

(c) By substituting g(X2) into the deformation gradient,

é ù
ê pX sin p (1 - X ) 1 p
cos (1 - X1 ) 0ú
ê 2
2 1
pX 2 2 ú
ê ú
ê p 1 p ú
F = êê- pX 2 cos (1 - X1 ) sin (1 - X1 ) 0úú
ê 2 pX 2 2 ú
ê ú
ê 0 0 1ú
ê ú
êë úû

The left Cauchy-Greed deformation tensor becomes

é pX 0 0ùú
ê 2
ê 1 ú
C = FT F = êê 0 0úú
ê pX 2 ú
ê 0 0 1úú
êë û
Note that the left Cauchy-Green deformation tensor has only diagonal components. Thus,
the matrices of eigenvectors and eigenvalues become
é pX 0 0ùú
ìï1 0 0ü ï ê 2
ïï ï
ï ê ú
1
F = í0 1 0ï
ï
ý, L = êê 0 0úú
ïï ï ê pX 2 ú
ïï0 0 1ï
ï ê 0
î ï
þ êë 0 1úú
û
Accordingly, U, Q, and V can be calculated by
é ù
ê pX 2 0 0ú
ê ú
T ê 1 ú
U = F LF = ê 0 0ú
ê pX 2 ú
ê ú
ê 0 0 1úú
êë û
é sin cos 0ù
ê ú
Q = FU-1 = êê- cos sin 0úú
ê ú
êë 0 0 1ú
û
 Index 13
é æ ö ù
ê 1 çç 1 ÷÷ ú
2
ê pX 2 sin + cos2 çç- pX 2 + ÷÷ sin cos 0ú
ê
ê pX 2 çè pX 2 ÷÷ø ú
ú
êæ ö ú
êçç 1 ÷÷ 1 ú
T
V = FQ = êç- pX 2 + ÷
÷÷ sin cos
2
pX 2 cos + sin 2
0ú
êçç p X ÷ø pX 2 ú
ê è 2 ú
ê 0 0 1úú
ê
ê ú
ê ú
ëê ûú

In the above expressions, simplified notations are used such that sin = sin p2 (1 - X1 ) and
cos = cos p2 (1 - X1 ) . ▄

P3.12 Consider a square element under pure shear deformation as shown in the figure.
The relation between deformed and undeformed coordinates becomes

x 1 = X 1 + kX 2 , x 2 = kX 1 + X 2

(a) Calculate deformation gradient F, Lagrangian strain E, Eulerian strain e, and

engineering strain e. (b) Calculate principal stretch tensors U and V, and rotation tensor
Q.
X2

X1

Solution:
(a) From the relation between deformed and undeformed coordinates, the deformation
gradient and Lagrangian strain can be calculated as
é1 k ù
F = êê ú
ú
êëk 1 úû
 é1 k2 k ùú
E = (F F - 1) = êê 2
1
2
T
1 2ú
êë k 2
k ú
û
The engineering strain becomes
é0 k ù
e = êê ú
ú
êëk 0 úû

Note that the Lagrangian strain has normal components in higher order.

(b) The relation in the polar decomposition is given as F = QU = VQ . First, the right
Cauchy-Green deformation tensor becomes
é1 + k 2 2k ùú
C = FT F = êê
êë 2k 1 + k 2 úú
û
The above Cauchy-Green tensor will have two eigenvalues:

C - l1 = (1 + k 2 - l)2 - 4k 2
= (1 + k 2 - l - 2k )(1 + k 2 - l + 2k ) = 0

Thus, two eigenvalues are l1 = (1 - k )2 and l2 = (1 + k )2 and

é(1 - k )2 0 ùú
L = êê
êë 0 (1 + k )2 úú
û
Two eigenvectors corresponding to the two eigenvalues can be calculated as
ì
ï ü ì ü é 1 1ù
E1 = ï1ï ï E2 =
ï
ï1ï
ï F= ê ú
1
í ý, 1
í ý, 1
ê-1 1ú
2 ï- 1ï 2 ï1ï 2
ï
î ï þ ï
îïþ ëê ûú
Then, the principal stretch tensor U can be calculated from the following relation:

1 éê 1 1ùú éê1 - k 0 ùú éê1 -1ùú éê1 k ùú

U = F LFT = =
2 êê-1 1úú êê 0 1 + k úú êê1 1 úú êêk 1úú
ë ûë ûë û ë û
Note that the principal stretch tensor is identical to the deformation gradient, which
means there is no rotation involved in the deformation. Thus, we have
é 1 0ù
Q = FU-1 = 1 = êê ú
ú
0 1
ëê úû

Because Q = 1 , the following relation can also be obtained:

V = FQ-1 = F = U
 Index 15
▄

P3.13 A square block of surface area A on all sides is under pure shear deformation due
to the uniformly distributed load F on the top surface, as shown in the figure. The
deformation of the block is such that the deformed coordinates can be written as

x 1 = X 1 + aX 2 , x 2 = X 2, x3 = X3

Calculate Cauchy stress, 1-st and 2-nd Piola-Kirchhoff stresses.

X2

X1

Solution:
Since the force is uniformly distributed over the area, the Cauchy stress will be
é 0 F / A 0ùú
ê
s = êêF / A 0 0úú
ê ú
êë 0 0 0ú
û
Using the relation of P = JF-1 s , the first Piola-Kirchhoff stress can be calculated by
é-aF / A F / A 0ù
ê ú
P = êê F / A 0 0úú
ê ú
êë 0 0 0ú
û
And from S = JF-1 sF-T the second Piola-Kirchhoff stress becomes
,
 é-2aF / A F / A 0ù
ê ú
ê
S= ê F /A 0 0úú
ê ú
êë 0 0 0ú
û
Note that the shear stress components are all the same, but negative normal stress
component exists in the case of P and S. However, this component becomes small as the
shear deformation becomes small. ▄

P3.14 A force R is applied at the tip of the uniform bar element shown in the figure. The
initial length and the cross-sectional area of the bar are, respectively, A0 and L0. The
elastic modulus of the material is E. Calculate the tip displacement by solving the total
Lagrangian variational equation with St. Vernant-Kirchhoff nonlinear elastic material
model. Assume the following numerical values: E = 700MPa, A0 = 1.0´10-4m2, L0 =
1.0m, and R = 10kN. Compare the tip displacement with that from the linear elastic
model when (a) E = 700MPa and (b) E = 70GPa.

1 2 R = 10kN

x
L0=1m

Solution:
If the tip displacement is uNL, the displacement in the bar can be approximated by

X
u(X ) = u = lX (1)
L0 NL

where l = uLN/L0 is the stretch ratio. Using Eq. (1), the displacement gradient can be
calculated by

du
0u = =l (2)
dX
Since the problem is 1D, the displacement gradient becomes a scalar. Note that the above
displacement gradient is in fact engineering strain. The deformation gradient becomes

L
F11 = 1 + 0u = 1 + l = (3)
L0

where L = L0 + uLN is the deformed length of the bar. The Lagrangian strain is given as

E11 = 21 (F11T F11 - 1) = 21 [(1 + l)2 - 1] = l + 21 l 2 (4)

 Index 17
Comparing with the engineering strain, the Lagrangian strain has an additional
quadratic term. Assuming that the material is St. Vernant-Kirchhoff nonlinear elastic, the
strain energy density becomes

W (E) = 12 E : D : E = 21 E ⋅ E112 = 12 E (l + 12 l 2 )2 (5)

By differentiating the above strain energy density with respect to Lagrangian strain, the
2nd Piola-Kirchhoff stress can be calculated as

dW
S11 = = E ⋅ E11 = E (l + 21 l 2 ) (6)
dE11

In order to obtain the variational equation, the variation of the strain needs to be
calculated. Since the displacement of the bar is expressed in terms of the tip displacement
in Eq. (1), the variation of the displacement can also be represented by that of the tip
displacement. Let uNL be the variation of the tip displacement, and l = uNL / L0 be the
variation of the stretch ratio, the variation of displacement and that of the Lagrangian
strain can be obtained as

X
u (X ) = u = lX (7)
L0 NL

E11 = 21 (F11T 0u + 0u T F11 ) = (1 + l)l (8)

Note that E11 is linear with respect to l . Using Eq. (8), the structural energy form for
the total Lagrangian formulation becomes
L0 L0
a 0 (u, u ) = ò 0
S11E11 A0 dX = ò 0
EA0 (l + 12 l 2 )(1 + l)l dX (9)

Since the integrand is independent of X, the integral can be evaluated analytically, as

a 0 (u, u ) = EA0L0l (l + 23 l 2 + 21 l 3 ) (10)

Since the point load is applied at the tip of the bar, the load form can be evaluated
without integration as

 0 (u ) = u (L0 )R = l L0R (11)

By equating Eqs. (10) and (11), the variational equation can be written as

EA0L0l (l + 23 l 2 + 21 l 3 ) = l L0R, "l Î  h (12)

In the discrete domain, the space of kinematically admissible displacements will be the
space of real numbers. In order to satisfy the above equation for all real number l , the
coefficients of l should be equal in the above equation, to yield

R
l + 23 l 2 + 21 l 3 = (13)
EA0

For linear elastic material, the above equation becomes

R
l= (14)
EA0

Note that both equations have the same right-hand side. Thus, the nonlinear displacement
will always be smaller than that of linear displacement. In addition, when the
displacement is small; i.e., l  1 , the higher-order terms can be negligible,
l 2 » l 3 » 0 . Thus, the nonlinear displacement will approach to the linear one.

(a) When E = 700 MPa, Eq. (13) yields l = 0.12028  uNL = 120.28mm , while
Eq. (14) yields l = 0.14286  uL = 142.86mm . Thus, the linear elastic model
predicts about 19% larger tip displacement.

(b) When E = 70 GPa, Eq. (13) yields l = 0.001426  uNL = 1.426mm , while Eq.
(14) yields l = 0.001429  uL = 1.429mm . Thus, the linear elastic model predicts
about 0.2% larger tip displacement.
▄

P3.15 Solve Problem P3.14 using force equilibrium; i.e., internal force caused by stress
is equal to external force.

Solution:
From the previous problem, we have

S11 = E (l + 12 l 2 )

However, the 2nd Piola-Kirchhoff stress cannot be used for force equilibrium. Thus, the
1st Piola-Kirchhoff stress is calculated using the following relation:

P11 = S11 ⋅ F11T = E (l + 12 l 2 )(1 + l) = E (l + 23 l 2 + 12 l 3 )

Note that the 1st Piola-Kirchhoff stress is defined with respect to the initial cross-
sectional area. Assuming this stress is uniform over the cross-section, the force
equilibrium can be obtained by

R = P11 ⋅ A0 = EA0 (l + 23 l 2 + 21 l 3 )

Thus, we obtain the same equation with the variational approach, as

 Index 19
R
l + 23 l 2 + 21 l 3 =
EA0

For linear elastic material, the above equation becomes

R
l=
EA0

Note that both equations have the same right-hand side. Thus, the nonlinear displacement
will always be smaller than that of linear displacement. In addition, when the
displacement is small; i.e., l  1 , the higher-order terms can be negligible,
l 2 » l 3 » 0 . Thus, the nonlinear displacement will approach to the linear one. ▄

P3.16 Consider a plane strain, unit depth, square element as shown in the figure. Use St.
Vernant-Kirchhoff isotropic material model with two Lame’s constants l and m. A
uniformly distributed force Tx (force per area) is horizontally applied at the top surface.
Assuming it is a simple shear problem, the deformation of the element can be written as
ìïx = X + kX
ï 1 1 2
í
ïïx 2 = X 2
î
(a) Find the relation between k and Tx , (b) Find the reaction force in X 2 direction at the
top surface, and (c) Compare the results with that of the linear elastic model.
X2

Tx

X1

Solution:
For the given deformation, the deformation gradient and Lagrangian strain can be
calculated as
 é1 k ù 1 1 é 0 k ùú
F = êê ú, E = (FT F - 1) = êê
ú 2 êk k úú
2
êë 0 1 úû 2
ë û
From the constitutive tensor of St. Vernant-Kirchhoff material C = l1 Ä 1 + 2mI and
the Lagrangian strain, the 2nd Piola-Kirchhoff stress can be calculated by
é 1 lk 2 mk ù
ê
S = D : E = l tr(E)1 + 2mE = ê 2 ú
2ú
êë mk ( 1
2
l + m)k úû

Since the 2nd Piola-Kirchhoff stress does not have any physical meaning, it is converted
to the 1st Piola-Kirchhoff stress as
é ( 1 l + m)k 2 mk ù
ê
T
P = S⋅F = ê 2 ú
3 2ú
êë mk + ( 2 l + m)k ( 2 l + m)k úû
1 1

The unit normal vector of the top surface is N = [0, 1]T. Thus, the surface traction on the
top surface becomes
ìïmk + ( 1 l + m)k 3 üï ìïT üï
PT ⋅ N = ïí
2 ïý = ïí x ïý
ïï ( 21 l + m)k 2 ïï ïïTy ïï
ïî ïþ î þ

(a) In the above equation, Tx is the horizontal surface traction on the top surface or,
equivalently, uniformly distributed force: Tx = mk + ( 21 l + m)k 3
(b) Ty is the uniformly distributed vertical reaction: Ty = ( 21 l + m)k 2
(c) For linear elastic model, the relation between k and Tx is Tx = mk . Thus, the
nonlinear model has higher-order terms. In addition, the vertical reaction in linear elastic
model is zero, while the nonlinear elastic model yields non-zero vertical reaction. When
k  0 , the results from nonlinear model approach that of the linear elastic model. ▄

P3.17 Consider a deformation of a rectangular bar whose deformed geometry is given as

x 1 = aX 1, x 2 = b X 2, x 3 = bX 3

When the material is incompressible and St. Venant-Kirchhoff material properties are
given as E = 600MPa and n = 0.49, write the expression of S11 component of the
second Piola-Kirchhoff stress as a function of a. In addition, write the expression of s11
of the Cauchy stress as a function of a. Plot S11 and s11 in the range of a = [0.7 1.5].

Solution:
For the given material properties, the Lame’s constants can be calculated from

En
l= = 9, 865.8 MPa
(1 + n )(1 - 2n )
E
m= = 201.3 MPa
2(1 + n )
 Index 21
For given deformation, the deformation gradient and Cauchy-Green deformation tensor
can be calculated as
éa 0 0 ù éa 2 0 0 ùú
ê ú ê
F = ê 0 b 0 úú ,
ê C = F F = êê 0
T
b2 0 úú
ê ú ê ú
êë 0 0 b úû ê0
ë
0 b2 ú
û
The relation between a and b can be obtained from incompressibility:

det F = ab 2 = 1  b = a-1/2

Thus, the Lagrangian strain can be calculated as

éa 2 - 1 0 0 ùú
ê
1 1
E = (FT F - 1) = êê 0 a-1 - 1 0 úú
2 2ê ú
ê 0 0 a-1 - 1ú
ë û
Since all shear components are zero, we can only consider the normal components as a
vector. The second Piola-Kirchhoff stress becomes
ìS ï
ï ü él + 2m l ùú ïïìE11 ïïü ìïï l2 (a + 2a - 3) + m(a - 1) ü ïï
2 -1 2
l
ïï 11 ï
ïï ê
ê l ï ï
ï ï ï ï
ï
S=ïíS 22 ý = ê l + 2m l úú ï 2 -1
íE22 ý = í l2 (a + 2a - 3) + m(a - 1)ý
-1
ï
ï ï
ï ê ú ï ï ï ïï
ï
ïS ï êë l l l + 2mú ï
ï E ï
ï ï
ï l
(a 2
+ 2a -1
- 3) + m(a -1
- 1) ïï
î 33 ï þ û ïî 33 ïþ ï î2 þ
The S11 component of the stress becomes

l 2
S11 = (a + 2a-1 - 3) + m(a2 - 1)
2
The Cauchy stress can be obtained from the relation

1
s= F ⋅ S ⋅ FT
J
The s11 component of the stress becomes

l 4
s11 = (a + 2a - 3a2 ) + m(a 4 - a2 )
2
The following figure shows the two stress components as a function of the principal
stretch a. Note that both stresses are highly nonlinear even if the relation between stress
and strain is constant.
 ▄

P3.18 Consider a simple shear deformation of a square whose deformed geometry is

given as

x 1 = X 1 + aX 2 , x 2 = X 2, x3 = X3

When the material is incompressible and St. Venant-Kirchhoff material properties are
given as E = 600MPa and n = 0.49, write the expression of S12 component of the
second Piola-Kirchhoff stress as a function of a. In addition, write the expression of s12
of the Cauchy stress as a function of a. Plot S12 and s12 in the range of a = [0.0 1.5].

Solution:
For the given material properties, the Lame’s constants can be calculated from

En
l= = 9, 865.8 MPa
(1 + n )(1 - 2n )
E
m= = 201.3 MPa
2(1 + n )

For given deformation with incompressibility, the deformation gradient and Lagrangian
strain tensor can be calculated as
é 1 a 0ù é0 a 0ùú
ê ú ê
1 T 1ê
F = ê0 1 0úú ,
ê E = (F F - 1) = êa a2 0úú
ê ú 2 2ê ú
êë0 0 1úû êë 0 0 1ú
û
In the plane strain problem, we can consider only three non-zero stress components. The
second Piola-Kirchhoff stress becomes
 Index 23
ì
ïïS ü ï él + 2m l 0 úù ì
ïï E ü ï ì
ïï l2 a2 üïï
11 ï ê 11 ï
ï
S=ï
ï
ï ê l ú ïï E ïï = ï( l + m)a2 ïï
ï
íS 22 ý = ê l + 2m 0 ú íï 22 ýï í ý
ï
ï ï
ï ê úï ï ï
ï
2
ï
ï
ï
ï 12 ï
S ï êë 0 0 mú ïï2E12 ïï ïï ma ï
î þ ûî þ î þï
Thus, S12 is a linear function of a. The Cauchy stress can be obtained from the relation

1
s= F ⋅ S ⋅ FT
J
The s12 component of the stress becomes
æl ö
s12 = çç + m÷÷÷ a 3 + ma
çè 2 ÷ø

Different from the hyperelastic material, now s12 is a cubic function, while S12 is a linear
function for the shear deformation. The following figure compares the two stresses as a
function of shear deformation.

P3.19 Consider the following deformation with ½a½ £ 1:

 x1 = X1 + aX 2, x 2 = 1 - a2 X 2 , x 3 = X3

Assume St. Venant-Kirchhoff material with two material parameters l and m. (a) Show
that the above deformation is a pure shear deformation in terms of the Lagrangian strain,
(b) Calculate the second Piola-Kirchhoff stress and Cauchy stress in terms of a, l, and m.

Solution:
(a) For the given deformation, the deformation gradient and the Lagrangian strain
become
é a 0úù é 0 a 0ù
ê1 ê ú
ê ú 1 T 1ê
F = ê0 1 - a2 0ú , E = (F F - 1) = êa 0 0úú
ê ú 2 2ê ú
ê0 0 1ú êë 0 0 0úû
ëê ûú
Thus, the given deformation is a pure shear deformation in terms of the Lgrangian strain.
(b) From the St. Venant-Kichhoff material, the stress-strain relation becomes
é 0 a 0ù
ê ú
S = l(trE)1 + 2mE = m êêa 0 0úú
ê ú
êë 0 0 0úû

Thus, only shear stress component S12 exists. The Cauchy stress becomes
é a2 ù
ê a 0 ú
ê 2 ú
ê 1-a ú
1 ê ú
s = FSFT = m êê a 0 0 ú
ú
J ê ú
ê ú
ê 0 0 0 ú
ê ú
ëê ûú
Note that the Cauchy stress has a non-zero normal component s11. The following figure
shows the deformation of a square block with a = 0.5.
 Index 25

P3.20 A force F is applied at the tip of the uniform bar shown in the figure. The
displacement of the bar is given as u(X) = lX where l is the principal stretch. The
initial length and the cross-sectional area of the bar are, respectively, A0 and L0. The
elastic modulus of the material is E. Calculate the tip displacement by solving the
principal stretch using the total Lagrangian formulation with the St. Venant-Kirchhoff
material model. Assume the following numerical values: E = 700 MPa, A0 = 1.0´10-4
m2, L0 = 1.0 m, and F = 10 kN. Compare the tip displacement with that from the linear
elastic model when (a) E = 700 MPa and (b) E = 70 GPa.

F = 10kN

x
L0=1m
Figure P3.20
Solution:
If the tip displacement is u NL , the displacement in the bar can be approximated by

X
u(X ) = u = lX
L0 NL
(1)
where l = u NL / L0 is the stretch ratio. Using Eq. (1), the displacement gradient can be
calculated by

du
0u = =l (2)
dX
Since the problem is 1D, the displacement gradient becomes a scalar. Note that the above
displacement gradient is in fact engineering strain. The deformation gradient becomes

L
F11 = 1 + 0u = 1 + l =
L0
(3)
where L = L 0 + u NL is the deformed length of the bar. The Lagrangian strain is given as
T
E11 = 21 (F11 F11 - 1) = 21 [(1 + l)2 - 1] = l + 21 l2
(4)
Comparing with the engineering strain, the Lagrangian strain has an additional quadratic
term. Assuming that the material is St. Vernant-Kirchhoff nonlinear elastic, the strain
energy density becomes
2
W (E) = 21 E : C : E = 21 E ⋅ E11 = 21 E (l + 21 l2 )2
(5)
By differentiating the above strain energy density with respect to Lagrangian strain, the
2nd Piola-Kirchhoff stress can be calculated as

dW
S11 = = E ⋅ E11 = E (l + 21 l 2 )
dE11
(6)
In order to obtain the variational equation, the variation of the strain needs to be
calculated. Since the displacement of the bar is expressed in terms of the tip displacement
in Eq. (1), the variation of the displacement can also be represented by that of the tip
displacement. Let u N L be the variation of the tip displacement, and l = u NL / L0 be the
variation of the stretch ratio, the variation of displacement and that of the Lagrangian
strain can be obtained as

X
u (X ) = u = lX
L0 NL
(7)
T
E11 = 21 (F11 0u + 0u T F11 ) = (1 + l)l
(8)
Note that E11 is linear with respect to l . Using Eq. (8), the structural energy form for the
total Lagrangian formulation becomes
L0 L0
a 0 (u, u ) = ò0 S11E11 A0 dX = ò0 EA0 (l + 21 l2 )(1 + l)l dX
(9)
Since the integrand is independent of X , the integral can be evaluated analytically, as
 Index 27
a0 (u, u ) = EA0L0l (l + 23 l2 + 21 l 3 )
(10)
Since the point load is applied at the tip of the bar, the load form can be evaluated
without integration as
 0 (u ) = u (L 0 )F = l L 0 F (11)

By equating Eqs. (10) and (11), the variational equation can be written as

EA0L0l (l + 23 l2 + 21 l 3 ) = l L0F, "l Î  h (12)

In the discrete domain, the space of kinematically admissible displacements will be the
space of real numbers. In order to satisfy the above equation for all real number l , the
coefficients of l should be equal in the above equation, to yield

F
l + 23 l 2 + 21 l 3 = (13)
EA0

For linear elastic material, the above equation becomes

F
l= (14)
EA0

Note that both equations have the same right-hand side. Thus, the nonlinear displacement
will always be smaller than that of linear displacement. In addition, when the
displacement is small; i.e., l  1 , the higher-order terms can be negligible,
l2 » l 3 » 0 . Thus, the nonlinear displacement will approach to the linear one.

(a) When E = 700 MPa, Eq. (13) yields l = 0.12028  uNL = 120.28mm , while
Eq. (14) yields l = 0.14286  uL = 142.86mm . Thus, the linear elastic model
predicts about 19% larger tip displacement.

(b) When E = 70 GPa, Eq. (13) yields l = 0.001426  uNL = 1.426mm , while
Eq. (14) yields l = 0.001429  uL = 1.429mm . Thus, the linear elastic model
predicts about 0.2% larger tip displacement. ▄

P3.21 Solve Problem P4.20 using force equilibrium; i.e., internal force caused by stress
is equal to external force.

Solution:
From the previous problem, we have
 S11 = E (l + 21 l2 )

However, the 2nd Piola-Kirchhoff stress cannot be used for force equilibrium. Thus, the
1st Piola-Kirchhoff stress is calculated using the following relation:
T
P11 = S11 ⋅ F11 = E (l + 21 l2 )(1 + l) = E (l + 23 l2 + 21 l 3 )

Note that the 1st Piola-Kirchhoff stress is defined with respect to the initial cross-
sectional area. Assuming this stress is uniform over the cross-section, the force
equilibrium can be obtained by

F = P11 ⋅ A0 = EA0 (l + 23 l2 + 21 l 3 )

Thus, we obtain the same equation with the variational approach, as

F
l + 23 l 2 + 21 l 3 =
EA0

For linear elastic material, the above equation becomes

F
l=
EA0

Note that both equations have the same right-hand side. Thus, the nonlinear displacement
will always be smaller than that of linear displacement. In addition, when the
displacement is small; i.e., l  1 , the higher-order terms can be negligible,
l2 » l 3 » 0 . Thus, the nonlinear displacement will approach to the linear one.
(a) When E = 700 MPa, Eq. (13) yields l = 0.12028  uNL = 120.28mm , while
Eq. (14) yields l = 0.14286  uL = 142.86mm . Thus, the linear elastic model
predicts about 19% larger tip displacement.

(b) When E = 70 GPa, Eq. (13) yields l = 0.001426  uNL = 1.426mm , while
Eq. (14) yields l = 0.001429  uL = 1.429mm . Thus, the linear elastic model
predicts about 0.2% larger tip displacement. ▄

P3.22 Consider two bar elements under a force at the tip. Using the displacement-
controlled method, plot the load-displacement curve (F vs. u2 and u3). Increase the tip
displacement u3 up to 1.0m with ten equal increments. Assume St. Venant-Kirchhoff
material with E = 100MPa, and cross-sectional areas of A(1) = 1.0´10-4 m2 and A(2) =
0.5´10-4 m2.
 Index 29

1 2 3
E, A(1) E, A(2) F

x
Figure P3.22
L0=1m L0=1m

Solution:
Since this is a one-dimensional problem, only X1-directional component will be used in
the following calculations. For Elements 1 and 2, the Lagrangian strains are defined as

E11(1) = u2 + 21 u22 , E11(2) = u3 - u2 + 12 (u3 - u2 )2

In the above equation u1 = 0 is used. For the St. Venant-Kirchhoff material, the second
Piola-Kirchhoff strsses for the two elements can be calculated by

S11(1) = E (u2 + 12 u22 ), (

S11(2) = E u 3 - u2 + 12 (u 3 - u2 )2 )
Since u1 is fixed, its variation is also equal to zero. The variation of the Lagrangian strans
become

E11(1) = (1 + u2 )u2, E11(2) = (u2 - u3 - 1)(u2 - u3 )

The energy form can be obtained by adding the contributions from two elements as
L0 2 L0
a(u, u ) = ò 0
S11(1)(1 + u2 )u2A(1) dX + ò
L0
S11(2)(u2 - u 3 - 1)(u2 - u 3 )A(2) dX
= u2 éêS11(1)(1 + u2 )A(1)L0 + S11(2)(u2 - u 3 - 1)A(2)L0 ùú - u 3 éêS11(2)(u2 - u 3 - 1)A(2)L0 ùú
ë û ë û
The load form is simply

(u ) = u3F3

Since the nonlinear variational equation must satisfy for arbitraty u2 and u3 , two
nonlinear equation can be obtained:

S11(1)(1 + u2 )A(1)L0 + S11(2)(u2 - u 3 - 1)A(2)L0 = 0

S11(2)(u2 - u3 - 1)A(2)L0 + F = 0

Note that the first equation is the coefficient of u2 and the second is that of u3 . Since u3
is prescribed for the displacement-controll method, its variation is zero, and the applied
force is in fact the reaction force required prescribing the displacement. Thus, for a given
u3, the unknown u2 is solved from the first equation. After that, the second equation is
used to solve for the reaction force F.
In order to solve the above nonlinear equation using the Newton-Raphson method,
the increment is stresses are required:

DS11(1) = E (1 + u2 )Du2, DS11(2) = E (u2 - u3 - 1)Du2

Note that only the increment Du2 is considered because the motion of u3 is prescribed;
i.e., the convergence iteration is performed to find Du2 after increasing u3 according to
the displacement controlled method. Thus,
L0

ò
a * (u; Du, u ) =
0
A(1)[E (1 + u2 )2 + S11(1) ]Du2u2 dX
2L0
+ò A(2)[E (u2 - u 3 - 1)2 + S11(2) ]Du2u2 dX
L0

( ) (
= u2L0 éêA(1) E (1 + u2 )2 + S11(1) + A(2) E (u2 - u 3 - 1)2 + S11(2)
ë )ùúû Du 2

Below is the list of MATLAB programs that solves for the nonlinear variational equation.
The following table and figure show the converged solutions at each increment.

%
% P3.22 Two bar elements--displacement controlled procedure
%
tol = 1.0e-5; conv = 0; u2 = 0; u2old = u2;
E = 1E8; A1 = 1E-4; A2 = .5E-4;
fprintf('\n step u2 u3 F');
% Displacement increment loop
for i=1:10
u3 = 0.1*i;
S1 = E*(u2+.5*u2^2);
S2 = E*(u3-u2+.5*(u3-u2)^2);
P = S1*A1*(1+u2)+S2*A2*(u2-u3-1);
F = S2*A2*(1+u3-u2);
R = -P;
conv = R^2;
% Convergence loop
iter = 0;
while conv > tol && iter < 50
Kt = A1*(E*(1+u2)^2+S1) + A2*(E*(u2-u3-1)^2+S2);
delu2 = R/Kt;
u2 = u2old + delu2;
S1 = E*(u2+.5*u2^2);
S2 = E*(u3-u2+.5*(u3-u2)^2);
P = S1*A1*(1+u2)+S2*A2*(u2-u3-1);
R = -P;
conv= R^2;
u2old = u2;
iter = iter + 1;
end
F = S2*A2*(1+u3-u2);
fprintf('\n %3d %7.5f %7.5f %8.3f',i,u2,u3,F);
end

Increment u2 u3 F
 Index 31
1 0.0343 0.1000 361.3
2 0.0704 0.2000 779.7
3 0.1077 0.3000 1256.9
4 0.1460 0.4000 1795.0
5 0.1851 0.5000 2396.4
6 0.2248 0.6000 3063.4
7 0.2651 0.7000 3798.5
8 0.3058 0.8000 4604.2
9 0.3469 0.9000 5482.9
10 0.3883 1.0000 6437.2

P3.23 Consider a nonlinear elastic uniaxial bar element under tip force F = 100N
shown in Figure 3.11. The stress strain relation is given in terms of Cauchy stress and
engineering strain in the deformed geometry: s11 = Ee11. Using the updated Lagrangian
formulation, solve for displacement at the tip, stress and strain of the uniaxial bar.
Assume E = 200Pa and the cross-sectional area A = 1.0m2.

Solution:
It is easy to estimate the stress and strain in order to make equilibrium with the applied
load. Since F = 100N and A = 1.0m2, the required Cauchy stress should be s11 =
100Pa. From the stress-strain relation, the required strain should be e11 = 0.5. Since the
strain is defined using the deformed geometry, the deformed length of the bar should be
2.0m, which yields the tip displacement of u2 = 1.0m. Now in the following, the updated
Lagrangian method is used to solve the problem.
Since the strain is defined in the deformed geometry,

du du d X u2 u2
e11(u ) = = = , e11(u ) = , J = 1 + u2
dx d X dx 1 + u2 1 + u2

Thus, since a concentrated force is applied at the tip, the load form is (u ) = Fu2 . The
energy form can be wriiten as
L

ò
a(u, u ) =
0
s e (u )A dx
11 11
L0
=ò s11e11(u )AJ dX
0
u2u2
= EAL0
1 + u2

Since the nonlinear variational equation satisfies for arbitrary u2 , the residual can be
defined as

u2
R = EAL0 -F = 0
1 + u2

Since the constitutive relation is given in terms of Cauchy stress and engineering strain, it
is unnecessary to transform the material description to the spatial descrition. It is more
convenient to directly linearlize the energy form in the spatial form. After linearizing the
residual, the incremental equation for the Newton-Raphson method becomes

EAL0
Du2 = -R
(1 + u2 )2

Below is the list of MATLAB programs that solve the the nonlinear variational equation.
Also the following table shows the convergence interation of the Newton-Raphson
method. The solution converges in the fourth iteration. As expected, the tip displacement,
strain, and strss converge to u2 = 1.0m., e11 = 0.5, and s11 = 100Pa.

%
% P3.23 Uniaxial bar--updated Lagrangian formulation
%
tol = 1.0e-5; iter = 0; E = 200;
u = 0; uold = u; f = 100;
strain = u/(1+u);
stress = E*strain;
P = stress;
R = f - P;
conv= R^2/(1+f^2);
fprintf('\n iter u1 Strain Stress conv');
fprintf('\n %3d %7.5f %7.5f %8.3f %12.3e %7.5f',iter,u,strain,stress,conv);
while conv > tol && iter < 20
Kt = E/(1+u)^2;
delu = R/Kt;
u = uold + delu;
strain = u/(1+u);
 Index 33
stress = E*strain;
P = stress;
R = f - P;
conv= R^2/(1+f^2);
uold = u;
iter = iter + 1;
fprintf('\n %3d %7.5f %7.5f %8.3f %12.3e %7.5f',iter,u,strain,stress,conv);
end

Iteration u Strain Stress conv

0 0.0000 0.0000 0.000 9.999E-01
1 0.5000 0.3333 66.667 1.111E-01
2 0.8750 0.4667 93.333 4.444E-03
3 0.9922 0.4980 99.608 1.538E-05
4 1.0000 0.5000 99.998 2.328E-10

P3.24 Consider a deformation of a rectangular bar whose deformed geometry is given as

x 1 = aX 1, x 2 = b X 2, x 3 = bX 3

When the material is incompressible, Mooney-Rivlin hyperelastic material with A10 =

80MPa and A01 = 20MPa, write the expression of S11 component of the second Piola-
Kirchhoff stress as a function of a. In addition, write the expression of s11 of the Cauchy
stress as a function of a. Plot S11 and s11 in the range of a = [0.7 1.5].

Solution:
For given deformation, the deformation gradient and Cauchy-Green deformation tensor
can be calculated as
éa 0 0 ù éa 2 0 0 ùú
ê ú ê
F = ê 0 b 0 úú ,
ê C = F F = êê 0
T
b2 0 úú
ê ú ê ú
êë 0 0 b úû ê0
ë
0 b2 ú
û
The relation between a and b can be obtained from incompressibility:

det F = ab 2 = 1  b = a-1/2

The three invariants of the deformation tensor can be obtained as

 I 1 = a2 - 2a-1
I 2 = 2a + a-2
I3 = 1

The reduced invariants become

J 1 = I 1I 3-1/3 = a 2 - 2a-1
J 2 = I 2I 3-2/3 = 2a + a-2
J 3 = I 3-1/2 = 1

In order to calculate stress, we need to differentiate the reduced invariants with respect to
strain
é 1 0 0ù
ê ú
I 1,E = 21 = 2 êê 0 1 0úú
ê ú
êë 0 0 1úû

é2a-1 0 0 ù
ê ú
I 2,E ê
= 2(I 1 1 - C) = 2 ê 0 2
a +a -1
0 ú
ú
ê 2 -1 ú
ê 0 0 a +a ú
ë û
éa-2 0 0 ùú
ê
I 3,E = 2I 3C-1 = 2 êê 0 a 0 úú
ê ú
êë 0 0 aú
û
The derivatives of the reduced invariants become
é2(1 - a-3 ) 0 0 ùú
ê
1 2
J 1,E = I 3-1/3I 1,E - I 1I 3-4/3I 3,E = êê 0 1 - a3 0 úú
3 3ê ú
ê 0 0 1 - a3 ú
ë û
é2(a-1 - a-4 ) 0 0 ù
ê ú
2 2ê ú
J 2,E = I 3-2/3I 2,E - I 2I 3-5/3I 3,E = ê 0 -a2 + a-1 0 ú
3 3ê 2 -1 ú
ê 0 0 -a + a ú
ë û
é a -2 00 ùú
ê
1
J 3,E = I 3-1/2I 3,E = êê 0 a 0 úú
2 ê ú
êë 0 0 aú
û
Thus, the second Piola-Kirchhoff stress becomes

S = A10J 1,E + A01J 2,E + K (J 3 - 1)J 3,E

 Index 35
The S11 component of the stress becomes

4é
S11 = ê A10 (1 - a-3 ) + A01(a-1 - a-4 )ùú
3 ë û
4
= (80 + 20a-1 )(1 - a-3 )
3
The Cauchy stress can be obtained from the relation

1
s= F ⋅ S ⋅ FT
J
The s11 component of the stress becomes

4
s11 = (80a2 + 20a)(1 - a-3 )
3
The following figure shows the two stress components as a function of the principal
stretch a. Note that the second Piola-Kirchhoff stress is highly nonlinear, but the Cauchy
stress is reasonably linear with respect to the principal stretch. Also note that the two
stresses are similar when the deformation is small. However, as deformation increases,
the difference also increases.

▄
P3.25 Consider a simple shear deformation of a square whose deformed geometry is
given as

x 1 = X 1 + aX 2 , x 2 = X 2, x3 = X3

When the material is incompressible, Mooney-Rivlin hyperelastic material with A10 =

80MPa and A01 = 20MPa, write the expression of S12 component of the second Piola-
Kirchhoff stress as a function of a. In addition, write the expression of s12 of the Cauchy
stress as a function of a. Plot S12 and s12 in the range of a = [0.0, 1.5].

Solution:
For given deformation, the deformation gradient and Cauchy-Green deformation tensor
can be calculated as
é 1 a 0ù é1 a 0ùú
ê ú ê
F = êê0 1 0úú , C = FT F = êêa a 2 + 1 0úú
ê ú ê ú
êë0 0 1úû êë 0 0 1ú
û
The three invariants of the deformation tensor can be obtained as

I 1 = a2 + 3, I 2 = a 2 + 3, I3 = 1

The reduced invariants become

J 1 = I 1I 3-1/3 = a2 + 3
J 2 = I 2I 3-2/3 = a2 + 3
J 3 = I 3-1/2 = 1

In order to calculate stress, we need to differentiate the reduced invariants with respect to
strain
é 1 0 0ù
ê ú
I 1,E = 21 = 2 êê 0 1 0úú
ê ú
êë 0 0 1úû

é2a2 + 4 -2a 0 ùú
ê
I 2,E = 2(I 1 1 - C) = êê -2a 4 0 úú
ê ú
ê 0 0 2a2 + 4ú
ë û
é2a2 + 2 -2a 0ù
ê ú
I 3,E = 2I 3C-1 = êê -2a 2 0úú
ê ú
êë 0 0 2ú
û
The derivatives of the reduced invariants become
 Index 37
é-a 4 - 4a2 3
a + 3a 0 ùú
ê
1 -4/3 2ê 3
J 1,E = I 3-1/3I 1,E - I 1I 3 I 3,E = ê a + 3a -a2 0 úú
3 3ê ú
ê 0 0 -a2 ú
ë û
é-2a 4 - 5a2 2a 3 + 3a 0 ùú
ê
2 2
J 2,E = I 3-2/3I 2,E - I 2I 3-5/3I 3,E = êê 2a 3 + 3a -2a2 0 úú
3 3ê ú
ê 0 0 -2a2 ú
ë û
éa2 + 1 -a 0ù
ê ú
1 -1/2 ê
J 3,E = I I 3,E = ê -a 1 0úú
2 3 ê ú
êë 0 0 1ú
û
Thus, the second Piola-Kirchhoff stress becomes

S = A10J 1,E + A01J 2,E + K (J 3 - 1)J 3,E

The S12 component of the stress becomes

2
S12 = [A10 (a 3 + 3a) + A01(2a 3 + 3a)]
3
= 80a 3 + 200a

The Cauchy stress can be obtained from the relation

1
s= F ⋅ S ⋅ FT
J
The s12 component of the stress becomes

s12 = 2(A10 + A01 )a = 200a

Note that the shear stress S12 is a cubic function of a, but s12 is a linear function.
 ▄

P3.26 Derive the energy form and its linearization of a Mooney-Rivlin hyperelastic
material using the perturbed Lagrangian method. Use a mixed variable r = [uT, p]T.

Solution:
Using the distortional strain energy density in Eq. (4.116) and dilatational strain energy
density in Eq. (4.123) for the Mooney-Rivlin material, the second Piola-Kirchhoff stress
can be obtained by

1 2
W (J 1, J 2 , J 3 , p) = A10 (J 1 - 1) + A01(J 2 - 3) + p(J 3 - 1) - p (1)
2K

S = W,E = A10J 1,E + A01J 2,E + pJ 3,E . (2)

Note that the independent pressure p is used. In order to derive the energy form, the first
variation of the strain energy density can be written as
æ pö
W = W,E : E + W,p p = S : E + çççJ 3 - 1 - ÷÷÷ p . (3)
è K ø÷

Since both displacement and pressure are independent variables, a new combined
variable is introduced as r = [uT, p]T. Then, the energy form can be obtained by
integrating Eq. (3) as

a(r, r ) º òò 0
W
S : E dW + òò 0 pH dW
W
(4)

where H = J3 – 1 – p/K corresponds to the volumetric strain.

The energy form a (r, r ) is nonlinear through the constitutive relation and strain-
displacement relation. Linearization of stress can be expressed in terms of displacement
and pressure increments as
 Index 39
DS = W,E,E : DE + W,E,p Dp = D : DE + J 3,EDp (4)

where D is the fourth-order constitutive tensor defined as

¶S
D= = A10J 1,EE + A01J 2,EE + pJ 3,EE (5)
¶E
and DE and Dp are the incremental strain and pressure. Linearization of energy form can
be obtained as
éE : (D : DE + J Dp) + S : DEù dW
a * (r; Dr, r ) º òòêë 0
W 3,E úû
æ Dp ÷ö (6)
+òò 0 p ççJ 3,E : DE - ÷ dW
W çè K ÷÷ø

The pressure term can be condensed on the finite element level by directly solving the
terms that contain the pressure variation. This can easily be done if constant pressure
approximation is used, which can be done within a finite element. ▄

P3.27 Derive the 6´6 [D] matrix in Eq. (3.147) for two-dimensional Mooney-Rivlin
material with three material parameters (A10, A01, and K). Use the penalty method for
near-incompressibility

Solution:
The expression for the constitutive tensor is given in Eq. (3.128). In order to make a
matrix expression, it is necessary to define the second-order derivatives of the three
invariants in the matrix notation as

[I 1,EE ] = [0]6´6

é0 4 4 0 0 ùú
0
ê
ê4 0 4 0 0 0 úú
ê
ê4 4 0 0 0 0 úú
ê
[I 2,EE ] = ê
ê0 0 0 -2 0 0 úú
ê ú
ê0 0 0 0 -2 0 ú
ê ú
ê0 0 0 0 0 -2ú
ë û
 é 4C 3 4C 2 -4C 5 ù
ê ú
ê 4C 4C - 4C ú
ê 3 1 6ú
ê 4C 4C 1 -4C 4 ú
[I 3,EE ] = êê 2 ú
ú
ê - 4C 4
- 2C 3
2C 6
2C 5 ú
ê ú
ê-4C 5 2C 6 -2C 1 2C 4 ú
ê ú
ê -4C 6 2C 5 2C 4 -2C 2 ú
ë û
From Eq. (3.129), the second-order derivatives of the three reduced invariants can be
written in the matrix notation as

1 -4 4 -7 1 -4
[J 1,EE ] = - I 3 3 ({I 1,E }{I 3,E }T + {I 3,E }{I 1,E }T ) + I 1I 3 3 {I 3,E }{I 3,E }T - I 1I 3 3 [I 3,EE ]
3 9 3
- 23 2 - 53 10 - 83 2 -5
[J 2,EE ] = I 3 [I 2,EE ] - I 3 ({I 2,E }{I 3,E } + {I 3,E }{I 2,E } ) + I 2I 3 {I 3,E }{I 3,E }T - I 2I 3 3 [I 3,EE ]
T T

3 9 3
1 - 23 1 - 1
[J 3,EE ] = - I 3 {I 3,E }{I 3,E }T + I 3 2 [I 3,EE ]
4 2

Note that the expressions of {I1,E}, {I2,E}, and {I3,E} are available in Section 3.5.2. From
Eq. (3.128), the constitutive matrix can be obtained as

[D] = A10 [J 1,EE ] + A01[J 2,EE ] + K (J 3 - 1)[J 3,EE ] + K {J 3,E }{J 3,E }T

P3.28 Derive the 6´6 [D] matrix in Eq. (3.147) for two-dimensional Mooney-Rivlin
material with three material parameters (A10, A01, and K). Use the perturbed Lagrangian
method for near-incompressibility

Solution:
For the perturbed Lagrangian formulation, the expression of stress is given in Example
3.15, and the constitutive tensor can be written as

¶S
D= = A10J 1,EE + A01J 2,EE + pJ 3,EE
¶E
In order to make a matrix expression, it is necessary to define the second-order
derivatives of the three invariants in the matrix notation as

[I 1,EE ] = [0]6´6
 Index 41
é0 4 4 0 0 0 ùú
ê
ê4 0 4 0 0 0 úú
ê
ê4 4 0 0 0 0 úú
ê
[I 2,EE ] = ê
ê0 0 0 -2 0 0 úú
ê ú
ê0 0 0 0 -2 0 ú
ê ú
ê0 0 0 0 0 -2ú
ë û
é 4C 3 4C 2 -4C 5 ù
ê ú
ê 4C 4C - 4C ú
ê 3 1 6ú
ê 4C 4C 1 -4C 4 ú
[I 3,EE ] = êê 2 ú
ú
ê - 4C 4
- 2C 3
2C 6
2C 5 ú
ê ú
ê-4C 5 2C 6 -2C 1 2C 4 ú
ê ú
ê -4C 6 2C 5 2C 4 -2C 2 ú
ë û
From Eq. (3.129), the second-order derivatives of the three reduced invariants can be
written in the matrix notation as

1 -4 4 -7 1 -4
[J 1,EE ] = - I 3 3 ({I 1,E }{I 3,E }T + {I 3,E }{I 1,E }T ) + I 1I 3 3 {I 3,E }{I 3,E }T - I 1I 3 3 [I 3,EE ]
3 9 3
- 23 2 - 53 10 - 8 2 -5
[J 2,EE ] = I 3 [I 2,EE ] - I 3 ({I 2,E }{I 3,E }T + {I 3,E }{I 2,E }T ) + I 2I 3 3 {I 3,E }{I 3,E }T - I 2I 3 3 [I 3,EE ]
3 9 3
1 - 23 1 - 1
[J 3,EE ] = - I 3 {I 3,E }{I 3,E }T + I 3 2 [I 3,EE ]
4 2

Note that the expressions of {I1,E}, {I2,E}, and {I3,E} are available in Section 3.5.2. Thus,
the constitutive matrix can be obtained as

[D] = A10 [J 1,EE ] + A01[J 2,EE ] + p[J 3,EE ]

P3.29 A nearly incompressible rubber block is confined between two frictionless rigid
walls as shown in the figure. When uniform pressure P is applied on the right end, the
length of the block is changed by x1 = (1 - a)X1. When a = 0.1, (a) calculate the value
of strain energy density and (b) the magnitude of applied pressure P on the right end.
Assume plane strain problem and use Mooney-Rivlin material with A10 = 80MPa, A01 =
20MPa, and K = 1,000MPa.
 Rigid
X1 Rubber P
wall Figure P3.29

Solution:
Since both X2 and X3 directions are fixed, the deformation of the rubber block can be
written as

x 1 = 0.9X 1, x 2 = X 2, x3 = X3

from which the deformation gradient and the right Cauchy-Green deformation tensor can
be calculated as
é.9 0 0ù é.81 0 0ù
ê ú ê ú
F = ê 0 1 0úú ,
ê C = ê 0 1 0úú
ê
ê ú ê ú
êë 0 0 1úû êë 0 0 1úû

The three eigenvalues of the right Cauchy-Green deformation tensor can be obtained as

l12 = l22 = 1, l32 = 0.81

Using the three eigenvalues, the three invariants can be obtained as

I 1 = 2.81, I 2 = 2.62, I 3 = 0.81

In addition, the reduced invariants are

J 1 = 3.0145, J 2 = 3.0151, J 3 = 0.9

(a) Strain energy density

K
W = A10 (J 1 - 3) + A01(J 2 - 3) + (J 3 - 1)2
2
= 80 ´ 0.0145 + 20 ´ 0.0151 + 500 ´ (-0.1)2
= 6.462
(b) Hydrostatic pressure

p = K (J 3 - 1) = -100MPa

P3.30 Consider a unit cube shown in Figure 3.15. Using an eight-node solid element,
perform biaxial extension analysis using ABAQUS. Apply uniform extensions in both X1
 Index 43
and X2 directions so that deformed shape will be 5´5´t3. Plot stress s11 and thickness
t3 as a function of stretch.

Solution:
The following program list shows the ABAQUS input file for the biaxial loading:

*HEADING *SOLID SECTION,ELSET=ONE,MATERIAL=MNEY

- INCOMPRESSIBLE HYPERELASTICITY *MATERIAL,NAME=MNEY
(MOONEY-RIVLIN), BIAXIAL TENSION *HYPERELASTIC, MOONEY-RIVLIN
*NODE,NSET=ALL 80., 20.,
1, *STEP,NLGEOM,INC=20
2,1. BIAXIAL TENSION
3,1.,1., *STATIC,DIRECT
4,0.,1., 1.,20.
5,0.,0.,1. *BOUNDARY,OP=NEW
6,1.,0.,1. FACE1,3
7,1.,1.,1. FACE3,2
8,0.,1.,1. FACE6,1
*NSET,NSET=FACE1 FACE4,1,1,5.
1,2,3,4 FACE5,2,2,5.
*NSET,NSET=FACE2 *EL PRINT,F=1
5,6,7,8 S,
*NSET,NSET=FACE3 E,
1,2,5,6 *NODE PRINT,F=1
*NSET,NSET=FACE4 U,RF
2,3,6,7 *OUTPUT,FIELD,FREQ=1
*NSET,NSET=FACE5 *ELEMENT OUTPUT
3,4,7,8 S,E
*NSET,NSET=FACE6 *OUTPUT,FIELD,FREQ=1
4,1,8,5 *NODE OUTPUT
*ELEMENT,TYPE=C3D8RH,ELSET=ONE U,RF
1,1,2,3,4,5,6,7,8 *END STEP

The analysis results are shown in the following figure:

▄