Vous êtes sur la page 1sur 27

Tensor, Stress, and Strain

Because we will adopt finite deformation kinematics to develop a finite element formulation based on thin shell theories, some background on tensor operations based on a
general coordinate system is needed and therefore is introduced briefly in this chapter.
For this chapter, we assume the reader has some background in vector and tensor analysis.
First, the concepts of stress and equilibrium are reviewed. The principal stresses and
stress invariants then are introduced. Because sheet metal forming involves large plastic
deformation, we then discuss the concepts of finite deformation kinematics. We also
discuss the small strain theory, which is widely used to describe deformation. Different
stress measures for finite element formulations are discussed. Finally, the stresses and
strains for tensile tests of sheet specimens are discussed.

El Transformation of Vectors and Tensors in Cartesian


Coor dinat e Systems

Figure 2.1 shows a set of rectangular Cartesian coordinates (x,y , z ) . We now change
the conventional coordinate notation from (x,y , z ) to (XI, x2,"3) to take advantage of the
indicia1 notation for vector and tensor operations. The base vectors for the coordinate
system (x,y , z ) are denoted as em ey and e, Now, the base vectors for the coordinate
system (XI, x2,"3) are denoted as el, e2, and e3. Note that these base vectors are orthogonal to each other, and the magnitudes of these base vectors are unity. We can express
the position vector x for point P a s
x = xe,+ye Y + ze,

(2.1.1)

based on the Cartesian coordinates (x,y , z). Alternatively, we can express the position
vector x for point P a s
3

x = xlel +x2e2 + x3e3 = C x i e i

(2.1.2)

i=l

based on the Cartesian coordinates (xl,x2,"3). We can omit the summation sign, and
the position vector x can now be written as

12

Mechanics Modeling of Sheet Metal Forming

Figure 2.1 A Cartesian


coordinate system.

The repeated subscript i is called the dummy index. The summation convention is
adopted for repeated indices. Usually, Latin indices such as iandjhave the range from
1 to 3 for three-dimensional problems, and Greek indices such as a andp have the range
of 1 to 2 for two-dimensional problems. When we mention the components of the position vector x, we can use xi where the subscript i can have the value of 1,2, or 3.
Figure 2.2 shows two sets of rectangular Cartesian coordinates (XI,x2,"3) and (xi,xi,xi)
with the same origin 0. The base vectors for the two coordinate systems are denoted
as ei and e> where the subscripts i and j can have the values of 1,2, or 3. The position
vector x can be expressed in terms of either of the two sets of the coordinates and the
corresponding base vectors as

To determine the component of x in the ej direction, we can use the dot product as

Here, Cjirepresent the Kronecker delta, which are defined as

6..
= 1 when
Jl

j =i

6ji= 0 when j

(2.1.6)
(2.1.7)

Tensor, Stress, and Strain

13

Figure 2.2 Two Cartesian


coordinate systems.

To determine the component of x in the


as

ej direction, we can also use the dot product

or

Here, aji(=
ej . ei) represent the direction cosines between the base vectors ej and
ei. In Eq. 2.1.9, the indexj is called the free index where the subscriptj can have the
value of 1,2, or 3. The index i again is called the dummy index because the summation
convention is implied for the repeated index i. The transformation of the coordinates
in Eq. 2.1.9 can be expressed in the matrix form as

(2.1.10)

Similarly,
(2.1.11)
or
(2.1.12)

14

Mechanics Modeling of Sheet Metal Forming

Equation 2.1.12 can be expressed in the matrix form as

(2.1.13)

where the superscript T means the transpose of the matrix. Note that occasionally
repeated indices do not mean summation if we specify "no sum on" (nso). For example,
the Ah diagonal element of a matrix C is represented by "Cjj nso i." For a rectangular
Cartesian coordinate system, the partial derivatives of Eqs. 2.1.9 and 2.1.12 can be
expressed in terms of aij as
(2.1.14)

Similarly, we consider a vector V, which can be expressed as

are the components of vector V in the ei direction, and vj' are the components
where l,$
of vector V in the ej direction. Similar to the discussion for the position vector x, we
can write
(2.1.16)

We may generalize the transformation rule to a tensor of the second order or higher. For
example, we consider a second order tensor W. The transformation rule is
(2.1.18)

(2.1.19)
Examples of second order tensors are the strain and stress tensors. Elastic modulus or
compliance tensors generally are expressed as fourth order tensors. In a rectangular
Cartesian coordinate system in this section, we may use either a subscript or a superscript to represent the components of a vector, a tensor, and a base vector, such as
V' 4,Wij Wij, and ez = ei. However, when a general coordinate system is used,
a subscript or superscript denotes the covariant or contravariant component of the field
quantity of interest.

Tensor, Stress, and Strain

15

E 2 Transformation of Vectors and Tensors in General


Coo rdinate Systems
A general coordinate system often is used to describe the geometry of a boundary value
problem in order to simplify the mathematical equations. For example, cylindrical and
spherical coordinate systems commonly are used. Transformation of the components of
a vector or tensor in one coordinate system to those in another system often is needed.
In sheet metal forming simulation, the computational formulation is based on the finite
strain thin shell theory, which is more conveniently expressed in terms of the curvilinear
coordinate system. However, when we obtain the computational results, we need to
express the results of the stress and strain measures in terms of the familiar Cartesian
coordinate system with a set of orthonormal base vectors. Therefore, in this section,
we present a summary of the transformation rule for tensor components between a
general coordinate system and a Cartesian coordinate system. Note again that it does
not matter whether a subscript or a superscript is used for the coordinates and the base
vectors in a Cartesian coordinate system. However, in a general coordinate system, a
subscript or superscript for the coordinates and the base vectors represents the covariant
or contravariant components [Spain, 19561. For example, we can express a position
vector x in a Cartesian coordinate system as
x = xi ei = xiei

(2.2.1)

For this case, ei = ei and xi = xi for the Cartesian coordinate system. Now we express
the position vector x in terms of a general coordinate system with a set of three base
vectors gl, g2, and g3, as shown in Figure 2.3. The general coordinate system has

o3\
\
\
\

x3

Figure 2.3 A Cartesian


coordinate system and a
general coordinate system.

16

Mechanics Modeling of Sheet Metal Forming

three coordinates 6'l, 6", and 6' 3, corresponding to the three base vectors gl, 82, and 83,
respectively. We call gl, g2, and g3 the covariant base vectors. Note that we use the
subscripts for the base vectors. The covariant base vectors gl, g2, and g3 do not have to
possess a magnitude of unity, and they do not have to be orthogonal to each other. The
position vector x can be expressed in terms of the two coordinate systems as
x = xiei = erg,

(2.2.2)

The covariant base vectors g, can be expressed as


(2.2.3)

We also can define the contravariant or the reciprocal base vectors gr


(2.2.4)

Note that unlike the base vectors e j and ei in the Cartesian coordinate system, the
covariant and contravariant base vectors g, and gr for the general coordinate system
are different. We will demonstrate this for a two-dimensional case later. When we use
the thin shell theory to describe the kinematics of curved sheet metals for sheet forming simulation and modeling, the base vectors g, and grin general are different point
by point and are functions of the location or the coordinates of the material point of
interest. We can easily show that the dot product of g, and g5 is
g, . gs = 6,"
where 6,"represent the Kronecker delta (6," = 1 when Y = s, and 6," = 0 when
Therefore, g, and gs are orthogonal to each other when r # s.

(2.2.5)
Y f s).

Now we can express a vector V referred to the covariant and contravariant base vectors g,
and g5 in a general coordinate system as
(2.2.6)

vs

c,

where
and are the contravariant and covariant components of vector V, respectively. Note that a bar is used to represent the component of the vector in the general
coordinate system. Figure 2.4 shows the vector V expressed in terms of the covariant
and contravariant base vectors in a two-dimensional plane. The figure shows three sets

of base vectors as (el,e2), (gl,g2), and g',g2 . Note that el is perpendicular to e2,
gl is perpendicular to g2, and g1 is perpendicular to 82. Note that the magnitudes of gl,
g2, gl, and g2are not equal to each other.
For a second order tensor W, we can express W in terms of the covariant and contravariant base vectors g, and g5 as
-

w = WYSg,gs= Wysg1'gS

(2.2.7)

Tensor, Stress, and Strain

17

Figure 2.4 A vector V


with respect to the
Cartesian, covariant, and
contravariant coordinate
systems as shown.

w"

q,,

where
and
are the contravariant and covariant components of W, respectively.
We usually use the covariant components to express a Lagrangian strain tensor and the
contravariant components to express a Cauchy stress tensor in a curvilinear coordinate
system when the sheets are under large deformation.

To examine the magnitudes of a vector and a tensor in a general coordinate system, we


need to use metric tensors. The metric tensor g,, is defined as
(2.2.8)

g, = g, ' g,

Here, g,, are symmetrical ( g , = g,,) due to the commutation of the dot product. In
general, all the terms of g,,are nonzeros for the set of the general base vectors gl, g2,
and g3. When g l , g2, and g3 are orthogonal to each other, only the diagonal terms survive when the metric tensor g,,is expressed in the matrix form. When g l , 82, and g3
are orthonormal (i.e., have the magnitudes of unity and are orthogonal to each other),
the metric tensor g,,reduces to 6,. The metric tensor 8" is defined as
gYs

g7'

. gs

(2.2.9)

Here, 8" also are symmetrical due to the commutation of the dot product. Based on the
definitions of the base vectors in Eqs. 2.2.3 and 2.2.4, we can easily show that
g7,qgq,= 6,s

(2.2.10)

Using the metric tensors, we can easily establish the relationship between the covariant
components ";. and the contravariant components
of vector V by the dot products as

vx

18

Mechanics Modeling of Sheet Metal Forming


-

K, = v . g , = g7J

(2.2.11)

(2.2.12)
Similarly, we can establish the relationship between the covariant components
the contravariant components @" of a tensor W as

@
, = g7,qgst@qt

@
, and

(2.2.13)

(2.2.14)
We can obtain the mixed components of a second order tensor that can be converted
from the covariant or the contravariant components using the metric tensors as

@
;

(2.2.15)

= gStFyt= g , , w S t

The magnitudes of the covariant and contravariant base vectors can be obtained from
the components of the metric tensors as

Ig,l = ( g , . g7,)1 1 2 = (g7,7,>' I 2


112
=

(Pi. . g r )

= (gYr)li2

nso r

(2.2.16)

nso r

(2.2.17)

We can convert the components of a vector V in a rectangular coordinate system to


those in a general coordinate system or vice versa. Taking the dot product of the base
vector ei,ei,g', or g , and V in Eq. 2.2.6 and using Eqs. 2.2.3 and 2.2.4, we can derive
the transformation laws for the components for the vector V as
(2.2.18)

v.= ,aer
-v,
ax
2

(2.2.19)

(2.2.20)

(2.2.21)
Note that 5 = V ifor the rectangular Cartesian coordinate system. Based on Eqs. 2.2.7,
2.2.3, and 2.2.4, we can derive the transformation laws for the components ofthe second
order tensor W as

Tensor, Stress, and Strain

19

(2.2.22)

(2.2.23)

(2.2.24)

(2.2.25)

In general, we use thin shell finite elements for sheet metal forming simulations. For a
thin shell, we use the middle surface as the reference. Although it initially is flat before
forming, the reference surface becomes curved during a forming process. We use a
set of curvilinear coordinates Oa (a = 1, 2) to represent the reference surface. This set
of coordinates also is called the Gaussian in geometric terms or convected coordinates
in mechanics terms. If the set is defined in a two-dimensional rectangular coordinate
system on an initially flat sheet, the metric tensor is the Kronecker delta. After deformation, the metric tensor on the deformed reference surface becomes aap (a,p = 1, 2),
of which the base vectors are not necessarily orthogonal and their magnitudes are not
necessarily unity.

I 2.3

Stress and Equilibrium

We consider a region of a solid with reference to a Cartesian coordinate system as


shown in Figure 2.5. The region has a volume Vand a surface S. As shown in this

TdS

Figure 2.5 A region of a


solid with a surface Sand
avolume I/:

20

Mechanics Modeling of Sheet Metal Forming

figure, x represents the position vector for point P, and v represents the velocity of
point P.Also, dsrepresents a differential element of the surface, and dvrepresents a
differential element of the volume. Here, n represents the outward unit normal to the
differential surface element dS, and T represents the force vector per unit area acting
on the differential surface element dS. Therefore, TdSrepresents the differential force
acting on the differential surface element dS. Likewise, f represents the body force per
unit volume, and fdvrepresents the differential force of the differential volume element
dl/: The linear momentum principle requires
(2.3.1)
where p represents the density and a represents the acceleration. Note that the momentum flux across the surface is not considered here.
The stress vector T represents the force per unit area on any surface in an average sense
from a continuum mechanics viewpoint. The orientation expressed in terms of the normal
of the surface element of interest is important. Consider a differential volume element
with respect to the Cartesian coordinate system, as shown in Figure 2.6. As shown in
this figure, we denote TI, T2, and T3as the tractions on the differential surfaces dSl, dS2,
and dS3, which have the outward normal in the xl, x2,and x3 directions, respectively.
Because TI, T2, and T3 are vectors, they can be decomposed into the components in the
XI,x2,and x3 directions, as shown in Figure 2.6, as

Figure 2.6 A differential


volume element with
respect to the Cartesian

r2
T1

(2.3.2)

(2.3.3)

T3 = 031el + 03Ze2 + 033e3

(2.3.4)

where el, e2, and e3 represent the unit vectors in the X I , x2,and x3 directions, respectively.
When we use the indicia1 notation, Eqs. 2.3.2 through 2.3.4 can be written as

Tensor, Stress, and Strain

21

where i represents the free index, which means that i can be assigned a value of 1, 2,
or 3 when the range of i is from 1 to 3. Here, j represents the dummy index where
the summation convention must be applied. For ov,
the first subscript i indicates the
direction of the normal to the differential area, and the second subscriptj indicates the
direction of the force acting on the differential area.
When the differential area dSis not perpendicular to the xl,x2, and x3 directions, we can
consider the equilibrium of the forces on a Cauchy tetrahedron as shown in Figure 2.7
to obtain the traction T on the inclined surface with the unit outward normal vector n.
In Figure 2.7, the tetrahedron has four differential surfaces: dSl, dS2, dS3, and dS.
Because the unit outward normal vectors for dS1, dS2, and dS3 are in the negative xl,
x2, and x3 directions, the tractions on these surfaces are denoted as -TI, -T2, and -T3,
based on the principle of action-reaction. With consideration of the linear momentum
equation for the tetrahedron as the size of the tetrahedron approaches zero, the volume
integrals approach zero faster than the surface integrals. Therefore, only the surface
integrals of traction over the four faces survive. Summing all the tractions over the four
faces leads to the traction vector T expressed in terms of the inner product of the unit
outward normal vector n and the stress tensor Q as

Figure 2.7 A tetrahedron


with the tractions T, -TI,
-T2, and -T3 on the differential surfaces dS, dSl,
dS2,and dS3, respectively, with the normals
in the n, -el, -e2, and -e3
directions.

T=n.a

(2.3.6)

where the unit outward normal vector n is expressed as

n = niei

(2.3.7)

and the stress tensor CT is expressed in the dyadic form as


Q

= Ojkejek

(2.3.8)

22

Mechanics Modeling of Sheet Metal Forming

In indicia1 notation, the component Tk of the traction vector T can be related to the
components oij of the stress tensor Q as
(2.3.9)

Tk = njojk

The linear momentum equation can be written in the component form as

I,T.dS + 1AdV =

paidV

(2.3.10)

Substituting Eq. 2.3.9 into Eq. 2.3.10 and using the divergence theorem, we can rewrite
Eq. 2.3.10 as

(2.3.11)

Because Eq. 2.3.1 1 must hold for any volume, the local equations of motion become

(2.3.12)

When we consider the angular moment equation, the stress tensor Q can be shown to
be symmetric as

Therefore, for three-dimensional solids where the indices range from 1 to 3, the stress
tensor Q has six components. When the contribution of the acceleration term pai is
small compared to the other terms, the local equations of motion are reduced to the
local equilibrium equations as

a oji
axj

-+h=O

(2.3.14)

For a typical sheet metal forming process, the contribution of the acceleration term is
quite small; therefore, quasi-static modeling based on the local equilibrium equations
is reasonable. When the magnitudes of the body force 4 are much smaller than those
involving the stress components oij,Eq. 2.3.14 becomes

(2.3.15)

Tensor, Stress, and Strain

23

p . 4 Principal Stresses and Stress Invariants


As shown in Figure 2.8, for a differential area dS with the unit normal vector n, the
direction of the traction vector T may not be in the same direction of the unit normal
vector n. We can express this in mathematical form as
T/IlTll

(2.4.1)

11

where the symbol


represents the magnitude ofthe vector quantity enclosed. However,
we can always find a normal vector n for a differential area dS such that the direction
of the traction vector T is the same as that of the unit normal vector n. Mathematically,
we can write
T/IlTIl =

(2.4.2)

Figure 2.8 A traction


vector T on a differential
area dS with the unit normal vector n. The direction of T may not be the
same as the direction of
the unit normal vector n.

This means that in this direction, only the normal stress exists, and no shear stress exists.
Because the traction vector T is in the same direction of the unit normal vector n, we
can write
T = n . o = oP n

(2.4.3)

Here, oprepresents the principal stress or the eigenvalue of the stress tensor Q. In the
component form,

where 6jk represent the Kronecker delta. Equation 2.4.4 can be written as
(2.4.5)

24

Mechanics Modeling of Sheet Metal Forming

Equation 2.4.5 represents three equations for the three components nj. Equation 2.4.5
can be expressed in the matrix form as

011 -

0-

012

021

022 -

O3 1

O32

O13

0-

O23

][i:]=o

(2.4.6)

O33 - Op

Because the components of the stress tensor


tions for nj exist when the determinant is

are real and symmetrical, nontrivial solu-

(r

(2.4.7)
or in the matrix form as

011 - 0 p

012

O13

O21

O22 - O p

O23

O31

O32

O33 - O p

=O

(2.4.8)

Equation 2.4.8 represents a cubic equation for op


3
-Op

+ I1Op2

- 120p

+ 13 = 0

(2.4.9)

Here, I,,12,and I3 are independent of the coordinate system chosen. They are called
the invariants of the stress tensor CT. I,,I,, and I3 are expressed as
1, = 0 k k = 0 1 1+ 0 2 2 + 0 3 3

(2.4.10)

(2.4.11)

(2.4.12)

where i i j k is the permutation symbol, which is defined as

1 when i, j , k are even permutations of 1 , 2 , 3


i i j k = -1 when i, j , k are odd permutations of 1 , 2 , 3
0 when any subscripts are repeated

(2.4.13)

Tensor, Stress, and Strain

25

Because CT is real and symmetric (oij= oji),


we will have three real roots (eigenvalues)
that are called the principal stresses. The corresponding directions of the eigenvectors
are the principal directions. In general, these directions are mutually orthogonal to each
other. Figure 2.9 shows a differential material element with the principal stresses oI,
oII,and oIIIcorresponding to the principal directions nI, nII, and nIII, respectively.

"3

1
"1

rc

Figure 2.9 A differential


material element with the
principal stresses oI,oII,
and oIIIcorresponding to
the principal directions nI,
"11, and "111, respectively.

nIII

2.5 Finite Deformation Kinematics

We consider a solid in motion, as shown in Figure 2.10. The reference configuration


represents the solid occupied at time t= 0. The current configuration represents the solid
occupied at time t. We consider a material point P a s shown in the figure. The material
point P a t time 0 has a position vector X, which defines the material coordinates Xl, X2,
and X3 for point P. The material point Pmoves to a new position P' at time t. Now the
material point P h a s a new position vector x, which defines the spatial coordinates xl,
x2,and x3 for the material point P. The spatial position vector x for the material point P
at the current time t can be expressed in terms of the reference position vector X as

x = x(X,t)

(2.5.1)

The displacement vector u for the material point P a t the current time t is defined as
u=x

x = x(X,t) - x

(2.5.2)

As shown in Figure 2.11, we now consider a differential material vector dX in the neighborhood of point P a t t= 0. At the current time t, the differential material vector dX is

26

Mechanics Modeling of Sheet Metal Forming

Reference configuration
(time 0)

\,Current configuration
\

Figure 2.10 A material


point P i n the reference
configuration at time 0 and
the current configuration at
time t.

(time t)

I
I
I
,fly-.

\
\

\
.-0

Time 0

0 / - - -

Timet

Figure 2.11 A differential


material vector dX and
the differential deformed
material vector a!x.

deformed into a!x. The magnitudes of dX and a%are defined as dSand ds, respectively,
as
dS = Jdx-dx =

ds =

,/a

4Z-X= 4
-

(2.5.3)

(2.5.4)

Tensor, Stress, and Strain

The stretch ratio A of the differential material vector dX is defined as

= ds/dS

(2.5.5)

The orientations of the material vectors dX and a% are defined as N and n, respectively,
as
=

dx/'Il4

n = dx/lldxll

(2.5.6)

(2.5.7)

The rotation of the material element dX can be obtained fi-om a comparison of N and n.
The deformation gradient tensor F is defined as

dx=F.dX

(2.5.8)

This equation can be written in the component form as


(2.5.9)
where
(2.5.10)

In this way, the deformation gradient tensor can be used to characterize the general
deformation near the material point P, for example, the stretch and rotation of the differential material vector dX.
The Lagrangian strain tensor E can be defined by

( d s ) 2- ( d S ) 2 = 2dX. E . dX

(2.5.11)

Substituting Eqs. 2.5.3 and 2.5.4 into Eq. 2.5.11 with use of Eq. 2.5.8 gives

Ex-F
2

.F-I

(2.5.12)

where FT represents the transposition of F, and I represents the unit tensor. In the
component form, Eq. 2.5.12 can be expressed as

27

28

Mechanics Modeling of Sheet Metal Forming

Equation 2.5.13 can be written as


(2.5.14)

The spatial coordinates xican be related to the material coordinates Xias

xi = xi

+ ui

(2.5.15)

Substituting Eq. 2.5.15 into Eq. 2.5.14 gives

(2.5.16)

When the magnitudes of the displacement gradients aui/&Yj are small, the Lagrangian
strains Ey are reduced to the small or infinitesimal strains cij as

(2.5.17)

k . 6 Small Strain Theory


Elementary definitions of strains can be illustrated based on simple deformation modes
of a differential rectangular material element with respect to the Cartesian coordinate, as shown in Figure 2.12. In this figure, the rectangular elements have two sides
AX1 and AX2 parallel to the xl and x2 directions. As shown in Figure 2.12(a), the
rectangular element is elongated in the x1 direction. The elongation of AX1 in the x1
direction is Au,. The strain sll can be estimated as

Figure 2.12 Elementary definitions of the


strains based on simple
deformation modes of a
differential rectangular
material element. (a) Tensile deformation in the
xl direction, (b) tensile
deformation in the x2
direction, and (c) shear
deformation with respect
to the xland x2 directions.

Tensor, Stress, and Strain

(2.6.1)

As shown in Figure 2.12(b), the rectangular element is elongated in the x2 direction.


2 be estimated as
The elongation of A X 2 in the x2 direction is Au2. The strain ~ 2 can
(2.6.2)

As shown in Figure 2.12(c), the rectangular material element is changed into a parallelogram. As shown in this figure, the rotation of AXl gives Au2, and the rotation of
AX2 gives Aul. The orientation changes for AX1 and AX2 are denoted as 01 and 62 as

(2.6.3)

(2.6.4)

In this figure, the angle between A X , and A X 2 decreases from n/2 to $. The engineering shear strain yl2 is defined as the decrease of the angle from n/2 to $, Therefore,
the engineering shear strain y12 can be estimated as

(2.6.5)

AS A X , and A X 2

+ 0 ,the definitions of ~ 1 1 , 1 2 2 and


,

are

(2.6.6)

(2.6.7)

az12 au,
ax, ax2

y12 = -+-

(2.6.8)

The shear strain for convenience of tensorial and mathematical operations is defined
as
(2.6.9)

29

30

Mechanics Modeling of Sheet Metal Forming

Figure 2.13 shows two neighboring material points P and Q and their displacements.
We now consider the small strain tensor from the viewpoint of the displacement gradient
tensor. Points Pand Q have the displacement vectors up and uQ,respectively, as
up = uiPei

(2.6.10)

= u?e.

(2.6.11)

UQ

A 2'

X1

Figure 2.13 Two neighboring material points P


and Q and their displacements.

X3

The relative displacement Au is defined as


Au = uQ - up

(2.6.12)

Au.z = uz Q - ui

(2.6.13)

In the component form,

The vector AX from P t o Q can be written as


AX = XQ - Xp

(2.6.14)
In the component form,
(2.6.15)

Tensor. Stress. and Strain

The components of the displacement gradient tensor Vu can be written as


(2.6.16)

To the first order approximation, we can write

(2.6.17)
P represent the components of the displacement gradient tensor at point P.
where ui,

As the magnitude of AX decreases to zero, Eq. 2.6.17 can be written as

We can decompose the displacement gradient tensor into a symmetric part and an antisymmetric part as
(2.6.19)

The first part or the symmetric part is defined as the infinitesimal strain tensor l i jas

(2.6.20)

The infinitesimal strain tensor E has six components. The infinitesimal strain tensor
defined here is consistent with the elementary definitions of the small strains from the
elongation and change of angle for two line elements in the two-dimensional cases, as
discussed earlier. The definition also is consistent with that of the Lagrangian strain
when the displacement gradients are small. The second part or the anti-symmetric part
as
is defined as the rotational tensor

(2.6.2 1)

The rotational tensor o has three components. The rotational tensor o can be shown to
relate to the rotation of a line element dX under the displacement field.

31

32

Mechanics Modeling of Sheet Metal Forming

Em
7 Different Stress Tensors
When a solid body is under finite deformation, different stress tensors are needed for
convenience. The Cauchy stress tensor Q is based on the force per unit deformed area.
Other stress tensors can be defined from the work rate per unit reference volume. The
work rate per unit reference volume can be written as

= t:D

(2.7.1)
= t:F
= S:E

where acolon ":" denotes the dyadic product. In Eq. 2.7.1, IF1 represents the determinant
of the deformation gradient tensor F, and D represents the rate of the deformation tensor, which is the symmetric part of the velocity gradient tensor Vu. Here, t represents
the Kirchhoff stress tensor, which can be defined as
t =

IFIQ

(2.7.2)

In Eq. 2.7.1, t represents the nominal or the first Piola-Kirchhoff stress tensor, which
is defined as
t =

IF~F-~

.Q

(2.7.3)

where F-' represents the inverse of the deformation gradient tensor F. In Eq. 2.7.1,
F is the deformation gradient rate tensor, which represents the work-conjugate strain
rate tensor to the nominal stress tensor t. Note that the nominal stress tensor t is not
symmetrical. In Eq. 2.7.1, S represents the second Piola-Kirchhoff stress tensor, which
is defined as

where ( F - l r represents the transpose of the inverse of the deformation gradient tensor

F-'. In Eq. 2.7.1, E is the Lagrangian strain rate tensor, which represents the workconjugate strain rate tensor to the second Piola-Kirchhoff stress tensor s. Note that the
second Piola-Kirchhoff stress tensor S is symmetric. These stresses are convenient to
use when a convected coordinate system is used to describe the kinematics.
Now we consider a solid body with reference to a Cartesian coordinate system, as
shown in Figure 2.14. Here, X represents the position vector for the material point P
in the reference or undeformed configuration, x represents the position vector for the
material point P i n the current or deformed configuration, and u represents the displacement vector for the material point P.The body has a volume Vo and a surface So in the

Tensor, Stress, and Strain

33

SO

T\

\
I

D'

Figure 2.14 A solid with


a material point P i n the
undeformed and deformed
configurations.

7 fdV

undeformed configuration. The body has a volume Vand a surface S i n the deformed
configuration. The classical principle of virtual work with respect to the deformed
configuration can written as

jsT@uidS

+ jV (A - pai)6uidV = jvoij6&ijdV

(2.7.5)

where 6ui represent the components of the virtual displacement 6u. Here, 6~~are

(2.7.6)

associated with the virtual displacement 6u. The principle ofvirtual work with respect
to the undeformed configuration can be written as

(2.7.7)

where $represent the components of the position vector X, and $represent the components of the first Piola-Kirchhoff stress tensor t. A subscript or superscript zero represents

34

Mechanics Modeling of Sheet Metal Forming

a field quantity for a material point in the undeformed configuration. In this case, the
equations of motion based on the undeformed configuration can be written as

(2.7.8)

The traction To referred to the undeformed surface with a unit normal N is related to
the first Piola-Kirchhoff stress tensor t as

When a solid body is under large deformation, a convected coordinate system can be
adopted to describe the kinematics. In a convected coordinate system, a set of coordinates (51,5 ~ ~ 5 3is)used to identify a material point in a solid in the undeformed and
deformed configurations. The position vectors of the material point of interest in the
undeformed and deformed configurations are denoted as
O x = O x ( 519

52 153)

= x(61$<Z3<3)

(2.7.10)
(2.7.11)

The displacement u is defined as

The displacement u usually is defined with respect to the base vectors associated with
the undeformed configuration. Because the sheet usually is flat before forming, a Cartesian coordinate system typically is used to describe the undeformed configuration. The
covariant base vectors in the undeformed and deformed configurations are defined as

(2.7.13)

dx

(2.7.14)

Usually, when a Cartesian coordinate system is used to describe the undeformed configuration, the base vectors ogi can be thought of as the usual Cartesian base vectors that
have a magnitude of unity and are mutually orthogonal to each other. Note that

Tensor, Stress, and Strain

The metric tensors associated with the undeformed and deformed bodies are

Ogij = Ogi ' Ogj

(2.7.16)

gij = g i ' g j

(2.7.17)

The covariant components of the Lagrangian strain E with respect to the base vectors
in the undeformed configuration are

(2.7.18)

The Lagrangian strain tensor E can be written in terms of the covariant derivatives of
the displacement u as

(2.7.19)

The stress tensors usually are defined with respect to the base vectors associated with
the deformed configuration, especially when plasticity theories are used to describe the
constitutive laws. Because

It easily can be shown that the contravariant components of the second PiolaKirchhoff stress S with respect to the undeformed base vectors are equal to the contravariant components of the Kirchhoff stress z with respect to the deformed base vectors

Finally, the classical principle of virtual work can be written alternatively with respect
to the undeformed configuration as

(2.7.22)
where

(2.7.23)

Note that the principle of virtual work is integrated over the original known configuration
of the solid body. This equation is the starting point for our finite element formulations.
Then the following step is to specify the constitutive relations between the Kirchhoff

35

36

Mechanics Modeling of Sheet Metal Forming

stress tensor T and the Lagrangian strain tensor E. Usually, a rate form of the principle of
virtual work is needed. Therefore, a rate form of the constitutive relation between 7 and
E is needed for setting up the finite element formulation. Commonly used constitutive
laws in the rate form for plastically deformed materials are discussed in Chapter 3.

p . 8 Stresses and Strains from Tensile Tests


Tensile tests typically are used to obtain the stress-strain relations for sheet metals. The
tensile stress-strain relation is used to extract the elastic modulus and the plastic strain
hardening relation. Figure 2.15 shows a typical sheet specimen. The middle portion of
the sheet specimen has the uniform cross-sectional area A. As shown in this figure, the
middle portion of the specimen is designated as the gage section with the gage length L.
The tensile stress-strain relation is extracted from the relation of the force and the elongation or extension of the gage length. When the reference gage length is marked as Lo
and the deformed gage length is marked as L, the engineering strain e is defined as

e x -L

- Lo

(2.8.1)

LO

F
"3

Figure 2.15 A typical


sheet specimen.

We can define the differential true strain d& as

dl
d& = 1

(2.8.2)

where I is the current gage length, and dl represents the differential gage length. The
true strain or the logarithmic strain E is defined as the differential strain integrated over
the deformation history from the reference gage length Lo to the current or deformed
gage length L as
(2.8.3)

Tensor, Stress, and Strain

37

Carrying out the integration gives the true strain E as


= In[$)

(2.8.4)

As the sheet specimen is extended, the cross-sectional area of the gage section decreases.
When the extension of the specimen is small due only to elastic deformation of the materials, the decrease in the cross-sectional area of the gage section is small and negligible.
The engineering stress S is defined as the force F applied to the specimen divided by
the original cross-sectional area A, as

s = -F

(2.8.5)

A0

When the extension of the bar becomes large due to plastic deformation, the decrease
of the cross-sectional area of the gage length becomes large and not negligible. The
true stress o is then defined as the force F divided by the deformed or current crosssectional area A as

F
o=-

(2.8.6)

When the large plastic strains are considered in the deformation history, the true stress
and true strain relation usually is used as the input of the materials. The engineering
stress-strain relation and the true stress-strain relation are plotted schematically in
Figure 2.16. In this figure, the initial part of the curve is linear, which represents the
linear elastic response of the material. The elastic portion is exaggerated in this figure.
As shown in Figure 2.16, the true stress-strain relation gives a higher stress for a given
strain when compared to the engineering stress-strain curve. For ductile sheet metals, the
engineering stress-strain curve usually shows a maximum, as shown in Figure 2.16.

Strain

p.9

Reference

Spain, B., Tensor Calculus, Oliver and Boyd, Edinburgh and London, U.K., 1956.

Figure 2.16 An engineering stress-strain curve and


the corresponding true
stress-strain curve.

Vous aimerez peut-être aussi