Sie sind auf Seite 1von 8

Notes on strain and deformation tensors

Marcel Frehner
September 2006
marcel.frehner@erdw.ethz.ch
for internal use only
These short notes summarize the basic principles and equations describing deformation and strain.
They do not contain a lot of words but mainly mathematical equations. Most of the descriptions are
taken from the internet site
www.engin.brown.edu/courses/en175/Notes/Kinematics/Kinematics.htm
which provides a more detailed description. A Matlab (Matlab 7.1.0.246) and a Maple (Maple 10)
script were written as well to understand the basic tensors described in this text. They are named
Matlab script: Strain_Tensors.m
Maple script: Strain_Tensors.mw
In the following text all bold letters denote vectors or tensors, normal letters denote scalar values.

1. Displacement field, displacement gradient tensor and transformation tensor

Definitions: P = (p1,p2)
Q = (q1,q2)
P = (p1,p2)
Q = (q1,q2)
l
l
m
m

therefore

dx
dx
Q
Q

=
=
=
=

Point before deformation with coordinates p1 and p2


Point before deformation with coordinates q1 and q2
Point after deformation with coordinates p1 and p2
Point after deformation with coordinates q1 and q2
length of line segment between points P and Q before deformation
length of line segment between points P and Q after deformation
vector of unit length pointing from P to Q before deformation
vector of unit length pointing from P to Q after deformation

lm
l m
P + dx
P + dx

vector connecting point P and Q


vector connecting point P and Q

The displacement field u(x,y) a vector field describes the displacement for every point. The
displacement gradient tensor H(x,y) describes the spatial change of the displacement field.

u x
u ( x, y ) =
u y

u x
x
H ( x, y ) =
u y

u x
y

u y

(1)

The displacement gradient tensor H can be used to calculate the displacement of point P and Q
assuming that points P and Q are infinitesimally close to the point of origin where the displacement
gradient tensor is defined here. With this assumption a first order Taylor expansion can be made using
the displacement gradient tensor.

P ' = P + u( p1 , p2 ) = P + HP = (I + H )P
Q ' = Q + u(q1 , q2 ) = Q + HQ = (I + H )Q = P + dx + HP + Hdx

(2)

where I is the identity tensor. The coordinate transformation written out:

u x
u x
1+

x
y p1
p1 '

=

u y p2
p2 ' u y
1+

x
y
144
42444
3

(3)

The transformation tensor F is defined as


F=I+H

(4)

The vector connecting the two points changes according to

du = dx '- dx = (Q '- P ') - (Q - P)


du = P + dx + HP + Hdx - P - HP - P - dx + P

(5)

du = Hdx
dx ' = dx + H dx = Fdx

or

(6)

F maps any undeformed vector into its deformed state (Eqation 6). This vector can also be a position
vector of a point. Therefore F also maps any point into its new position after deformation (Equation 3).
According to Equation (6) the deformation gradient tensor maps a undeformed vector into its deformed
state. Considering two successive deformations Equation (6) is applied twice.

dx ' = F1dx
dx '' = F2dx ' = F2 F1dx = Ftot dx

(7)

where dx represents the vector after two deformation steps. The finite transformation tensor Ftot maps
the undeformed vector dx directly into its state after two deformation steps dx. This procedure can be
applied to any number of deformation steps. The finite transformation tensor Ftot then maps the
undeformed vector directly into its final state. All following considerations can be applied to F, as well
as to Ftot, except formulations for infinitesimal strain.

2. Cauchy-Green strain tensors

Using Equation (6) the length of a vector after the deformation l can be expressed with the length
before the deformation l.

l '2 = dx 'T dx ' = (Fdx)T Fdx = dxT FT Fdx = l 2mT FT Fm

(8)

Quadratic extension of a vector is defined as follows. Equation (8) is substituted into the definition.

l '2 l 2 mT FT Fm
T T
F Fm
=
=m{
2
2
l
l
C

(9)

where C is the right Cauchy-Green strain tensor.

C = FT F

(10)

Using Equation (6) the length of a vector before the deformation l can be expressed with the length
after the deformation l.

l 2 = dxT dx = (F 1dx ')T F 1dx ' = dx 'T F T F 1dx ' = l '2 m 'T ( FFT ) m '
1

(11)

Inverse Quadratic extension I of a vector is defined as follows. Equation (11) is substituted into the
definition.
2
T
T
1
l 2 l ' m ' ( FF ) m '
I = 2 =
= m 'T ( FFT ) m '
2
123
l'
l'
1

(12)

where B is the left Cauchy-Green strain tensor.

(13)

B = FFT

Both Cauchy-Green strain tensors contain information about the strain, i.e. change of length of a
vector. They are both symmetric and positive definite. They do not contain information about the rigid
body rotation, i.e. rotation of a vector without change of length. The information of the total
deformation, i.e. change of length of a vector AND rigid body rotation is only provided by the
displacement gradient tensor H or by the transformation tensor F.

3. Lagrange strain tensor

The Lagrange description of strain is similar to the Cauchy-Green description of the quadratic strain
(Equation 9). It only uses a different definition of the quadratic extension, here called the Lagrange
quadratic extension L. Equation (8) is again substituted into the definition.
2
T T
l '2 l 2 l 2mT FT Fm l 2 l ( m F Fm 1)
=
=
2l 2
2l 2
2l 2
mT FT Fm 1
1
L =
= mT ( FT F I ) m
2
2 243
14

L =

(14)

The Lagrange strain tensor E is defined as


E=

1 T
(F F I)
2

(15)

4. Eulerian strain tensor


The Eulerian description of strain is similar to the Cauchy-Green description of the inverse quadratic
strain (Equation 12). It only uses a different definition of the inverse quadratic extension, here called
the Eulerian quadratic extension E. Equation (11) is again substituted into the definition.

2
T
T
2
2
T
T 1
l '2 l 2 l ' l ' m ' ( FF ) m ' l ' 1 m ' ( FF ) m '
=
=
E =
2l '2
2l '2
2l '2

1 m 'T ( FFT ) m '


1

E =

1
1
= m'
I ( FFT ) m '
2 42443
14
T

(16)

E*

The Eulerian strain tensor E* is defined as


E* =

1
1
I ( FFT )
2

(17)

Both Lagrange and Eulerian strain tensors contain information about the strain, i.e. change of length of
a vector. They are both symmetric and positive definite. They do not contain information about the
rigid body rotation, i.e. rotation of a vector without change of length. The information of the total
deformation, i.e. change of length of a vector AND rigid body rotation is only provided by the
displacement gradient tensor H or by the transformation tensor F.

5. Infinitesimal strain tensor and infinitesimal rotation tensor


The general Lagrange strain tensor E written out is

u x
1 +
1 T
1
x
E = (F F I) =
2
2 u x
y

u y u x
1 + x
x
u y u y
1+

y x

2
2

u u u
2 x + x + y
x x x
1
E=
2 u
u x u x u y u y u y
x
y + x y + x + y x

u x

y
1 0

u y 0 1
1+

u x u x u x u y u y u y

+
+
+
y
x y
x
y x

2
2
u y u y u x

2
+
+

y y y

(18)

For infinitesimal strain, i.e. all terms are much smaller than 1, the quadratic terms in Equation (18) are
negligible. The infinitesimal strain tensor results

u x
2
x
1
E=
2 u x u y
+

x
y
=

This is equal to

u x u y
+

y
x
u y
2

1
( H + HT )
2

(19)

(20)

The infinitesimal strain tensor is only an approximation of the more general Lagrange strain tensor for
small strains. It contains information about the strain, i.e. change of length of a vector. It is symmetric
and positive definite. It does not contain information about the rigid body rotation, i.e. rotation of a
vector without change of length. The information of the total deformation, i.e. change of length of a
vector AND rigid body rotation is only provided by the displacement gradient tensor H or by the
transformation tensor F.
Equivalent to the infinitesimal strain tensor the infinitesimal rotation tensor can be defined. It is
also an approximation for a more general rotation tensor R for small rotations. It describes the rigid
body rotation, i.e. rotation of a vector without change of length. R is not described here.
=

1
( H HT )
2

(21)

Only the combination of the infinitesimal strain tensor and the infinitesimal rotation tensor
provides information about the total deformation.

H =+

(22)

6. Principal values and directions of strain


There are several equivalent ways to calculate the principal stretching values e1 and e2.
1. With the eigenvalues C of the right Cauchy-Green strain tensor C
c
e1,2
= C 1

(23)

2. With the eigenvalues B of the left Cauchy-Green strain tensor B


b
e1,2
= B 1

(24)

3. With the eigenvalues of the Lagrange strain tensor E


e
e1,2
= 1 + 2 E 1

(25)

4. With the eigenvalues E* or the Eulerian strain tensor E*


e*
e1,2
= 1 2 E * 1

(26)

5. With the eigenvalues of the infinitesimal strain tensor . This method provides only an
approximation for small strains.

e1,2

(27)

The principal stretching values 1 and 2 describe the stretching of the two vectors that are stretched
maximally and minimally in a certain displacement field. The length change of these vectors is

l ' = (1 + e1,2 ) l
l'
= (1 + e1 )
l max

or

l'
= (1 + e2 )
l min

(28)

(29)

The orientation of these two vectors can either be defined before or after the deformation.
1. Before deformation: The orientation of a vector that WILL BE stretched maximally or
minimally in a certain displacement field is given by

the normalized eigenvectors vC1 and vC2 of the right Cauchy-Green strain tensor C
the normalized eigenvectors vE1 and vE2 of the Lagrange strain tensor E
the normalized eigenvectors v1 and v2 of the infinitesimal strain tensor . These
eigenvectors are an approximation for small strains.

2. After deformation: The orientation of a vector that WAS stretched maximally or minimally
in a certain displacement field is given by

the normalized eigenvectors vB1 and vB2 of the left Cauchy-Green strain tensor B
the normalized eigenvectors vE*1 and vE*2 of the Eulerian strain tensor E*

7. Strain ellipse

The strain ellipse is a graphical display of the state of strain. Before deformation a circle is assumed
which is then deformed in a certain displacement field. The resulting ellipse shows the direction and
length of the two principal stretching directions. So the vectors of the major and minor axes of the
strain ellipse are

a major = (1 + e1C ) v1C = (1 + e1E ) v1E (1 + e1 ) v1


a minor = (1 + e2B ) v 2B = (1 + e2E * ) v 2E *

(30)

A finite strain ellipse, i.e. a deformed circle after several deformation steps can be calculated the same
way but using the finite transformation tensor Ftot according to Equation (7).

8. Some helpful tensor rules


There are some tensor rules that may help with the handling of tensors.

A ( B + C ) = AB + AC
A ( BC ) = ( AB ) C

( AB )

= BT A T

(A ) = (A )
( A B) = B
1 T

( AB )

= A T
A T

= B 1 A 1

Das könnte Ihnen auch gefallen