38 CHAPTER 3.

KINEMATICS

3.1. Given the two-dimensional body

B = {(X1 , X2 ) | 0.1 < X1 < 1 , 0.1 < X2 < 1},

and the displacement field

u1 = u · e1 = 0.2 ∗ ln(1 + X1 + X2 ), (3.1)

u2 = u · e2 = 0.2 ∗ exp(X1 ), (3.2)

plot the displaced shape of the body.

Do not do this problem by hand! Use Matlab.
The plots of the reference and deformed configurations should look like those in Fig. 3.1.

1.4

1.2

0.8

0.6

0.4

0.2

-0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4

Figure 3.1: Undeformed configuration — dotted lines, and deformed confinguration — solid lines.

Hint. First set up some characteristic lines on the body. For each point compute its displacement
and add it to the original position by noting xi = ei · (X + u).

Solution 3.1. The script file used to make these plots

function []= LogExpMap()
% Define characteristic lines for plotting

c1 = [ linspace(.1,1,50)’ 0.1*ones(50,1) ];
c2 = [ linspace(.1,1,50)’ 0.55*ones(50,1) ];
c3 = [ linspace(.1,1,50)’ 1.0*ones(50,1) ];
c4 = [ 0.1*ones(50,1) linspace(.1,1,50)’ ];
c5 = [ 0.55*ones(50,1) linspace(.1,1,50)’ ];
c6 = [ 1.0*ones(50,1) linspace(.1,1,50)’ ];

% Plot the original body

plot(c1(:,1),c1(:,2),’:’,’Linewidth’,2); hold on
plot(c2(:,1),c2(:,2),’:’,’Linewidth’,2);
plot(c3(:,1),c3(:,2),’:’,’Linewidth’,2);
plot(c4(:,1),c4(:,2),’:’,’Linewidth’,2);
plot(c5(:,1),c5(:,2),’:’,’Linewidth’,2);
plot(c6(:,1),c6(:,2),’:’,’Linewidth’,2);
 39

% Set aspect ratio

axis equal

%Add motion to initial points and replot

c1 = addmotion(c1);
c2 = addmotion(c2);
c3 = addmotion(c3);
c4 = addmotion(c4);
c5 = addmotion(c5);
c6 = addmotion(c6);

% Replot on figure

plot(c1(:,1),c1(:,2),’Linewidth’,2);
plot(c2(:,1),c2(:,2),’Linewidth’,2);
plot(c3(:,1),c3(:,2),’Linewidth’,2);
plot(c4(:,1),c4(:,2),’Linewidth’,2);
plot(c5(:,1),c5(:,2),’Linewidth’,2);
plot(c6(:,1),c6(:,2),’Linewidth’,2);
axis equal
hold off

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function [xy] = addmotion(c)

% function to add motion to array of (x,y) pairs

xy(:,1) = c(:,1) + 0.2*log(1 + c(:,1) + c(:,2));

xy(:,2) = c(:,2) + 0.2*exp(c(:,1));

 45

3.4. Given the matrix of components of the deformation gradient tensor F with respect to an orthonormal
basis {ei },  
3 −1 0
[F] = 2 2 0 .
0 0 1
Using MATLAB, calculate:
(a) The principal stretches and the principal stretch directions which are the eigenvalues and eigen-
vectors of the right stretch tensor U.
(b) The matrices of U, R.
(c) Green-Strain E = (1/2)(C − 1).
(d) The Hencky or logarithmic strain
X
ln U = ln λi ri ⊗ ri ,
i

where λi and {ri } are the eigenvalues and eigenvectors of the stretch tensor U.
(e) Comment on the differences (if any) in the values of the components of E and ln U.

Solution 3.4. (a)-(d) We use the following function in MATLAB to obtain the solution:

function [] = problem3( )

% Define the Deformation Gradient

F = [3 -1 0; 2 2 0;0 0 1];

% Define the Right Cauchy Green Tensor

C = F’*F;

%Calculate the eigenvalues and eignevectors of C

[vectors, values] = eig(C);

% select the eigenvalues from "values"

omega1 = values(1,1);
omega2 = values(2,2);
omega3 = values(3,3);

% Calculate the principal stretches

lambda1=sqrt(omega1);
lambda2=sqrt(omega2);
lambda3=sqrt(omega3);

% Select the columns of vectors as the eigenvectors

r1 = vectors(:,1);
r2 = vectors(:,2);
r3 = vectors(:,3);

% Construct U in the principal basis

U(1,1) = lambda1;
U(2,2) = lambda2;
U(3,3) = lambda3;
46 CHAPTER 3. KINEMATICS

% Map the matrix back to the original coordinates

U = vectors*U*vectors’;

% The rotation tensor R is

R = F*inv(U);

% Green Strain
E_g = (1/2)*(C - eye(3));

% To find the Hencky strain we need the logarithm of U so we use the same
% procedure as we did for taking the square root of C
[eigenvectors, eigenvalues] = eig(U);
E_h(1,1) = log(lambda1);
E_h(2,2) = log(lambda2);
E_h(3,3) = log(lambda3);
E_h = vectors*E_h*vectors’;

%Display Results
disp(’Right Cauchy-Green Tensor C’); C
disp(’first principal stretch’); lambda1
disp(’second principal stretch’); lambda2
disp(’third principal stretch’); lambda3
disp(’first principal stretch direction’); r1
disp(’second principal stretch direction’); r2
disp(’third principal stretch direction’); r3
disp(’Stretch Tensor’); U
disp(’Rotation Tensor’); R
disp(’Green-Strain’); E_g
disp(’Hencky Strain’); E_h

The output of this code is:

Right Cauchy-Green Tensor C

C =

13 1 0
1 5 0
0 0 1

first principal stretch

lambda1 =

second principal stretch

lambda2 =

2.2084

third principal stretch

 47

lambda3 =

3.6226

first principal stretch direction

r1 =

0
0
1

second principal stretch direction

r2 =

0.1222
-0.9925
0

third principal stretch direction

r3 =

-0.9925
-0.1222
0

Stretch Tensor

U =

3.6015 0.1715 0
0.1715 2.2295 0
0 0 1.0000

Rotation Tensor

R =

0.8575 -0.5145 0
0.5145 0.8575 0
0 0 1.0000

Green-Strain

E_g =

6.0000 0.5000 0
0.5000 2.0000 0
0 0 0
48 CHAPTER 3. KINEMATICS

Hencky Strain

E_h =

1.2798 0.0600 0
0.0600 0.7996 0
0 0 0

(e) The values of the strain components of E and those of ln U are very different from each
other at such large deformations.
56 CHAPTER 3. KINEMATICS

3.9. High resolution transmission electron microscopy (HRTEM) is an experimental method that allows
one to image materials down to sub-atomic level resolution. When applied to crystalline materials
(say metals) it allows one to image the location of the atoms accurately. Consider the two schematic
representations in Fig. 3.5 of what one can typically find when applied to a metal sample at two
different temperatures. Upon cooling from 350 K to 320 K the material has undergone a martensitic

10.1 deg

26.6 deg

25.0 deg
m
0n
0.5

0.45 nm
m
1n
0.4
0.34 nm

T = 350K T = 320K

Figure 3.6: Schematic of a two-dimensional diffusion-less lattice transformation.

(diffusion-less) transformation from one set of atomic spacings and angles to another. Determine the
“transformation” stretch-tensor associated with the transformation shown. Treat as a two-dimensional
problem; note, the diagram is not to scale. Figure 3.6 is an actual HRTEM image just so you can have
an appreciation of what is experimentally possible. The white dots are the atoms! The scale bar at
the top right is only 1 nm.

Figure 3.7: HRTEM image of Gum Metal. The white dots are atoms.
 57

Solution 3.9. In this problem one can identify two vectors and how they map from the reference
configuration to the spatial configuration. This provides sufficient information
√ to
√ compute all 4
components of the deformation gradient. Knowing F, one computes U as C = FT · F.
One can identify two starting vectors:

V(1) = 0.34e1
(2)
V = 0.45[cos(63.4o )e1 + sin(63.4o )e2 ] = 0.2015e1 + 0.4024e2

These two map respectively to

v(1) = 0.41[cos(25.0o )e1 + sin(25.0o )e2 ] = 0.3716e1 + 0.1733e2

(2)
v = 0.50[cos(79.9o )e1 + sin(79.9o )e2 ] = 0.0877e1 + 0.4923e2

From the relation that v(1) = F · V(1) one finds that F11 = 1.0929 and F21 = 0.5096. Then from
v(2) = F · V(2) one finds that F12 = −0.3294 and F22 = 0.9682. This then gives
 
1.4542 0.1334
C∼
0.1334 1.0459

Compute the eigenvalues and vectors of C gives:

 
0.2855
(λ1 )2 = 1.0061 r1 ∼
−0.9584

and  
2 −0.9584
(λ2 ) = 1.4939 r2 ∼ .
−0.2855
Thus the right stretch tensor is given as:
 
1.2044 0.0600
U = λ1 r1 ⊗ r1 + λ2 r2 ⊗ r2 ∼ .
0.0600 1.0209

Compact MATLAB code to perform the computations:

VV = [ 0.34 0.45*cosd(63.4);
0.00 0.45*sind(63.4) ];
vv = [0.41*cosd(25.0) 0.50*cosd(79.9);
0.41*sind(25.0) 0.50*sind(79.9) ];
F = vv*inv(VV)
C = F’*F
[evec,eval] = eig(C)
U = sqrt(eval(1,1))*evec(:,1)*evec(:,1)’ + ...
sqrt(eval(2,2))*evec(:,2)*evec(:,2)’
 61

3.11. Consider the deformation map x1 = X1 , x2 = X2 +kX12 , x3 = X3 for a unit cube B = [0, 1]×[0, 1]×[0, 1],
where k is a scalar parameter.
(a) Compute the normal strain at ( 12 , 12 , 12 ) in the direction oriented along the vector (1, 2, 3). [Warn-
ing! Normalize!]
√ √
(b) Compute√ the √
engineering shear at this point between the two direction n = (1/ 2, 1/ 2, 0) and
m(−1/ 2, 1/ 2, 0).
(c) For parts 3.11a and 3.11b, determine the value below which k has to be for the small strain
approximations to have a relative error of less than 10−4 . For the case of engineering shear, also
examine the absolute error.

Solution 3.11. For this problem first compute F and then C. Evaluate C at ( 12 , 12 , 12 ). This
gives:  
1 0 0
[F] =  2kX1 1 0 
0 0 1
and
1 + 4k 2 X12
    
1 2kX1 0 1 0 0 2kX1 0
[C] = [F> · F] =  0 1 0   2kX1 1 0 = 2kX1 1 0 .
0 0 1 0 0 1 0 0 1

At ( 21 , 12 , 12 ) this gives:
1 + k2
 
k 0
[C] =  k 1 0 .
0 0 1
√
(a) Denoting the direction as e, eA ∼ (1, 2, 3)/ 14, one has
r
p 14 + 4k + k 2
λ(e) = ei Cij ej − 1 = −1
14

(b) Likewise, one can evaluate the engineering shear relation to give:

−k 2
   
−1 n · Cm −1
γ = sin = sin √ √
λ(n)λ(m) 2 + 2k + k 2 2 − 2k + k 2
−k 2
 
−1
= sin √
4 + k4

(c) Noting that  

  0 k/2 0
1
[] = (∇u + (∇u)> ) =  k/2 0 0 ,
2
0 0 0
the linear normal strain is e · e = k/7. The linear part of the orthogonal shear is n · 2m = 0.
Comparing to the full nonlinear expressions shows that for the linear normal strains to have
a relative error less than 10−4 one needs k / 5.6 × 10−4 . For the engineering shears relative
error is always 1 if k 6= 0. However for the absolute error to be less than 10−4 one needs
k / 1.4 × 10−2 .
62 CHAPTER 3. KINEMATICS

3.12. Consider a square bar B = [− a2 , a2 ] × [− a2 , a2 ] × [0, L], of side length a and longitudinal length L. The
bar is deformed according to

χ(X) = (X1 + βX2 + αX3 ) e1 + (X2 − βX1 + αX3 ) e2 + X3 + αX32 e3 ,

(3.5)

where α and β are given constants.

(a) Determine the deformation gradient; assume arbitrary α and β.
(b) Determine the Green strain tensor; assume arbitrary α and β.
(c) Under the assumption that |α|  1 and that |β|  1, determine the small strain tensor.

Solution 3.12. Fij = ∂χi /∂Xj so:

 
1 β α
[F] =  −β 1 α 
0 0 1 + 2αX3

The Green strain is E = 12 (F> F − 1):

    
1 −β 0 1 β α 1 0 0
1    −β 1 − 0
[E] = β 1 0 α 1 0 
2
α α 1 + 2αX3 0 0 1 + 2αX3 0 0 1
β2
 
0 α(1 − β)
1
=  0 β2 α(1 + β) 
2
α(1 − β) α(1 + β) α2 (1 + 4X32 ) + 4αX3

The small strain tensor is  = 21 (∇u + ∇u> )

   
0 β α 0 −β 0
1 
[] = −β 0 α + β 0 0 
2
0 0 2αX3 α α 2αX3
 
0 0 α
1
=  0 0 α 
2
α α 4αX3

One observes that if one ignores all terms of second order and higher, α2 , αβ, β 2 , then E reduces
to .