MMK Bab 5

12/20/2017 Stress Introduction
Search Continuum Mechanics Website

Stress Introduction Search
home > stress > stress intro
Introduction
This page reviews the familiar stress tensor. Stress
is always simply F orce/Area , but some
complexity does arrise because the relative
orientation of the force vector to the surface normal
dictates the type of stress. When the force vector is
normal to the surface, as shown to the right, the
stress is called normal stress and represented by .
When the force vector is parallel to the surface, the

stress is called shear stress and represented by .
When the force vector is somewhere in between,
then its normal and parallel components are used as
follows.
Fnormal Fparallel
= and =
Ref: http://en.wikipedia.org/wiki/File:Axial_stress.svg
A A
This page is a near-duplicate of the earlier stress.html page in the Introductory Mechanics section. If
you have read that page, then this one can be skipped.
The earlier page served as the complete discussion of stress because it was in the
Introductory Mechanics section (keyword here being Introductory). This time, this page is only the
introduction to a full chapter on the subject. The reason being that, this time, we will worry about the
complications that arise when large deformations and rotations are present.
For example, when large deformations are present, does one use the initial or deformed area to
calculate stress? And if a part rotates 90 such that a force originally in the x-direction ends up acting
in the y-direction, then should the corresponding stress be xx or yy ?
Component Definitions
Many components are needed to capture a

http://www.continuummechanics.org/stressintroduction.html 1/5
y p p
complete stress state. Consider the object in
the 2-D example here that is being pulled in F
simple tension, though not in a direction
parallel to any global axis. Standard practice
is to (virtually) cut it perpendicular to the
global axes as shown. This first cut results in
an area with unit normal parallel to the global
F
x-axis. The force on this area contains both
normal and parallel components. The
F
stresses are defined as y Fy
Fx Fy
Fx
xx =
Ax
and xy =
Ax
x F
Note how the two subscripts on the stress variables match those on the force and area components with one
subscript coming from each.
Alternately, one could (virtually) cut the object horizontally to produce a surface with an outward normal in the y-
direction. This leads to
Fy Fx
yy = and yx =
Ay Ay
If a numerical example were worked out, one would notice an amazing result. It is that xy = yx . This will
always be true in order to maintain rotational equilibrium. This is discussed in more detail next.
Equilibrium
The complete (2D) stress state at a point is shown below. The key difference between the left and right figures
is the shear stresses. But they will be discussed later.
First, let's look at the normal stresses, xx and yy . Note how the x-normal stress, xx , is present on both the
left and right sides of each square in order to maintain horizontal equilibrium. These x-normal stresses
represent tension because they point out of the square. Tensile normal stresses have positive values, and
compressive normal stresses have negative values.
The y-normal stresses, yy , are also present on two surfaces, top and bottom, in order to maintain vertical
equilibrium. Like xx , yy is also drawn to represent tension, which is positive.
The difference between the left and right pictures is that yx in the left figure is replaced by xy in the right
figure. The left figure contains two shear stress values, xy , which rotates the square counter-clockwise, and yx
, which rotates the square clockwise. But if the two shear values are not equal, then the square will not be in
rotational equilibrium. The only way to maintain rotational equilibrium is for xy to be equal to yx . So there is no
need to have two separate variables. The right figure contains only one, xy .
syy syy
t
http://www.continuummechanics.org/stressintroduction.html
t 2/5
tyx txy
yy yy
txy Rotational txy

sxx y
Equilibrium sxx y
txy x sxx x sxx

txy
tyx txy
syy syy
2-D Notation
Stress is in fact a tensor. Why? Because it obeys standard coordinate transformation principles of tensors. This
alone appears to be enough to make it so. It can be written in any of several different forms as follows. They are
all identical.
11 12 xx xy xx xy
= [ ] = [ ] = [ ]
21 22 yx yy yx yy
But since xy = yx , all the tensors can also be written as
11 12 xx xy xx xy
= [ ] = [ ] = [ ]
12 22 xy yy xy yy
Setting xy = yx has the effect of making (requiring in fact) the stress tensors symmetric.
3-D Notation
All of the above conventions in 2-D also apply to the 3-D case. Notation for the 3-D case is as follows.
11 12 13 xx xy xz xx xy xz

= 21 22 23 = yx yy yz = yx yy yz

31 32 33 zx zy zz zx zy zz
But rotational equilibrium requires that xy = yx , xz = zx , and yz = zy . This also produces symmetric
tensors.
Ref: http://commons.wikimedia.org/wiki/File:Stress_in_a_continuum.svg
11 12 13 xx xy xz xx xy xz

= 22 23 = xy yy yz = xy yy yz
12

13 23 33 xz yz zz xz yz zz
0 Comments Continuum Mechanics

1 Login
Sort by Oldest
Recommend Share
Start the discussion
LOG IN WITH
OR SIGN UP WITH DISQUS ?
Name
Be the first to comment.
Subscribe d Add Disqus to your siteAdd DisqusAdd Privacy
Thank You Stress Chapter - $4.99 Entire Website - $12.95
Thank you for visiting this For $4.99, you receive two For $12.95, you receive two
webpage. Feel free to email formatted PDFs (the first for formatted PDFs (the first for
me if you have questions. 8.5" x 11" pages, the second 8.5" x 11" pages, the second
for tablets) of the complete for tablets) of the entire
Also, please consider visiting stress chapter. website.
an advertiser above. Doing so
g
helps to cover website hosting Click here to see a sample Click here to see a sample
fees. page in each of the two page in each of the two
formats. formats.
Bob McGinty
bmcginty@gmail.com Email address to receive PDFs Email address to receive PDFs
Copyright 2012 by Bob McGinty
Table of Contents
Stress Traction Vector
12/20/2017 Traction Vectors

Traction Vectors Search
home > stress > traction vector
Introduction
The traction vector, T, is simply the force
vector on a cross-section divided by that
cross-section's area.
F
T =
Area
So T has units of stress, like MPa, but it is

absolutely a vector, not a stress tensor. So all
the usual rules for vectors apply to it. For Ref: http://commons.wikimedia.org/wiki/File:stress_vector.svg
example, dot products, cross products, and
coordinate transforms can be applied.
Calculating a Traction Vector
The object below has a 400 mm2 cross sectional area and is being pulled in tension by a
4,000 N force (red) in the x-direction. So the (arbitrarily chosen) rightward pointing internal force
vector (blue) is
F = 4, 000 i N
y
F F F
x
It is cut (virtually) so the traction vector is
http://www.continuummechanics.org/tractionvector.html 1/9
1
T = ( ) 4,000 i N = 10.0 i MPa
2
400 mm
Note the units are MPa now rather than N.
This time, the object is cut (virtually again) at a 30 angle.
y
F F F
x
The applicable area in this case is
2
400 mm
2
A = = 462 mm

cos(30 )
And the traction vector is
1
T = ( ) 4,000 i N = 8.66 i MPa
2
462 mm
Note the direction of the traction vector is always the same as the internal force vector. Only its
magnitude changes with cut angle.
Normal and Shear Stresses

Normal and shear stresses are simply the components of the traction vector that are normal and parallel to
the area's surface as shown in the figure. Using n for the unit normal vector to the surface, and s for the
unit vector parallel to it, means that
= T n and = T s
It's very important to recognize that and here are each scalar values, not full tensors. This is the
natural result of the dot product operations involving T, n, and s . (Dot products produce scalar results.)
The normal and shear stress values here are scalars rather than tensors because they are only two
individual components of the full stress tensor.
s
n
Also, note that in 3-D, there are in fact an infinite
number of s vectors parallel to the surface each
12/20/2017
number of s vectors parallel to the surface, each
Traction Vectors
n
having a different component in-and-out of the
page, so to speak. This is why it is common to
specify one parallel to the page and a second
T
perpendicular to it.
t s
Normal and Shear Stress from a Traction Vector
Recall that the traction vector from the above example was
1
T = 4,000 i N = 8.66 i MPa
2
462 mm
The unit normal to the surface is

n = (cos 30 , sin 30 , 0)
So the normal stress on the surface is

= T n = (8.66, 0, 0) (cos 30 , sin 30 , 0) = 7.5 MPa
The vector parallel to the surface is

s = ( sin 30 , cos 30 , 0)
The shear stress on the surface is

= T s = (8.66, 0, 0) ( sin 30 , cos 30 , 0) = 4.33 MPa
Stress Tensors and Traction Vectors

The relationship between the traction vector and
stress state at a point results directly from setting A cosO A
http://www.continuummechanics.org/tractionvector.html T 3/9
12/20/2017
O
Traction Vectors
A
the sum of forces on an object equal to zero, i.e.,
imposing equilibrium.
T
O
xx A cos + xy A sin = Tx A sxx Ty

y
xy A cos + yy A sin = Ty A
Tx
x
The area, A, cancels out of both sides leaving
txy
xx cos + xy sin = Tx
syy
xy cos + yy sin = Ty
A sinO
but cos and sin are the components of the unit normal to the surface, n = (cos , sin ) , that T is
acting on.
Replacing the cos and sin with n x and n y gives
xx n x + xy n y = Tx T
xy n x + yy n y = Ty
O
Ty
Both equations can be summarized as
sxx Tx
y
T = n n = (cosO , sin O)
x
or in tensor notation as
txy
Ti = ij n j
syy
The above equations are very useful, compact,
matrix and tensor notation representations of the
equilibrium equations. The full equations, in 3-D, are
xx n x + xy n y + xz n z = Tx
yx n x + yy n y + yz n z = Ty
zx n x + zy n y + zz n z = Tz
The tensor notation term, ij n j , leads to nine separate stress components. For example, both xz and
zx are present above, and both are always equal. This is in fact common in all equations involving stress
and strain.
Traction Vector from Stress Tensor

Given the stress tensor (in MPa)
50 10 30

= 10 95 20

30 20 15
Calculate the traction vector on a surface with unit normal n .

= (0.400, 0.600, 0.693)
Tx 50 10 30 0.400 46.79

Ty = 10 95 20 0.600 = 74.86

Tz 30 20 15 0.693 34.40
So T = 46.79 i + 74.86 j + 34.40 k MPa .
If the area is 100 mm2, then the force on it would be F = 4, 679 i + 7, 486 j + 3, 440 k N .
Stress Transforms
This section introduces an aspect of coordinate transformations of stress tensors that is a subset of the
general case, which comes later. It does so by combining different equations involving the traction vector.
Recall that the normal and shear stresses on a surface are related to the traction vector by
= T n and = T s
Recall that the normal and shear stresses here are just scalar quantities on the surface, not a full stress
tensor.
But we also saw that the traction vector is related to the full stress tensor by
T = n
Substituting this equation for T into the above ones gives
= n n and = s n
In tensor notation, the equations are
= ij n i n j and = ij si n j
These represent very useful relationships between the stress tensor in the global coordinate system and
the normal and shear stress components at any other orientation.
Stress Transform Example
Recall the above stress tensor
50 10 30

= 10 95 20

30 20 15
We had calculated the traction vector on a surface with unit normal

n = (0.400, 0.600, 0.693). This time, calculate the normal and shear stresses on this
surface.
The normal stress on the surface is
{0.400 0.600 0.693} 50 10 30 0.400

= 10 95 20 0.600

30 20 15 0.693
= 87.47MPa
In order to compute a shear stress, we first need a specific one of an infinite number of unit
vectors parallel to the surface. Let's choose s = (0.832, 0.555, 0.000). A dot product will
verify that this vector is perpendicular to n.
{0.832 0.555 0.000} 50 10 30 0.400

= 10 95 20 0.600

30 20 15 0.693
= 2.62MPa
So there is very little shear on this face in the given s direction. But this doesn't mean that there
is no shear on the face at all. To see this, choose a second direction parallel to the surface and
perpendicular to the first s. Obtain this by crossing the unit normal vector with the first tangential
vector.
n s = (0.400 i + 0.600 j + 0.693 k) (0.832 i + 0.555 j + 0.000 k)
= 0.385 i 0.576 j + 0.721 k
So the shear in the direction perpendicular to the first is

So the shear in the direction perpendicular to the first is
{-0.385 -0.576 0.721} 50 10 30 0.400

= 10 95 20 0.600

30 20 15 0.693
= 36.33MPa
So there is a good bit of shear stress in this perpendicular direction. And the negative value
indicates that it is in the direction opposite of the s direction.
Transformation Tip
This transformation "trick" could be used to compute the normal and shear stresses on all six
faces of a cube at any random orientation, and in the process, perform a complete coordinate
transformation of a stress tensor. But it's actually easier to do = Q Q just as is the
T
case for strain tensors.
But the opposite is also true. The stress tensor can be replaced with the strain tensor to obtain
normal = n n and /2 = s n
Or in tensor notation as
normal = ij n i n j and /2 = ij si n j
This works because since both stress and strain are tensors, then any math operation that
applies to one also applies to the other.
For a quick math review, note that si n j in the above equations can be interpreted as a diadic product of
the two vectors, s n. And then this result is "double dotted" with the stress or strain tensor to obtain the
final scalar shear value. So the calculation could be written as
= s n = : (s n)
(The same could also be done to compute the normal stress as well.)
This diadic product for shear arises so often in metal plasticity that it is represented by the single letter p,
and named the Schmidt tensor after the engineer who studied metal plasticity in the early 1900's.
s1 n 1 s1 n 2 s1 n 3

p = s n = s2 n 1 s2 n 2 s2 n 3


s3 n 1 s3 n 2 s3 n 3
Forces on Cross Sections

Before closing, recall one more time that the force on a
cross-section is
F = T dA = n dA
Both forms turn up often in literature.

1 Login
Sort by Oldest
Recommend Share
Join the discussion
LOG IN WITH
Name
B j hK
Brajesh Kumar 3 months ago
Sir i have confusion in this sentence."And the negative value indicates that it is in the direction opposite
of the s direction." What will be the opposite direction since the two directions of shear stress are
perpendicular.
Reply Share
BobMcG Moderator > Brajesh Kumar 3 months ago

Thanks for the interest, Brajesh. The "s" is a vector with direction: 0.385i0.576j+0.721k. The
fact that this led to a negative shear value (tau = -36.33 MPa) means that the net shear force on
the face is in a direction opposite to the "s" vector direction. Or in other words, the shear stress,
resulted from a shear force in the opposite direction.
Reply Share
Click here to see a sample Click here to see a sample
helps to cover website
page in each of the two page in each of the two
hosting fees.
formats. formats.
Bob McGinty
Email address to receive PDFs Email address to receive PDFs
bmcginty@gmail.com
Table of Contents
Stress Introduction Energetic Conjugates
12/20/2017 Energetic Conjugates

Energetic Conjugates Search
home > stress > energetic conjugates
Introduction
This page introduces several definitions of stress. A key discriminator among the different stress tensors is
whether they report stress in a material's undeformed, and especially unrotated state, (the reference
configuration), or in its deformed and rotated state, (the current configuration).
It is interesting that most, perhaps even all, stress definitions can be paired with a corresponding strain
tensor. They come in pairs such that the product of the two will give strain energy, hence the name of this
page. This does not mean that the corresponding pairs must be used together when performing structural
analyses. But they must be when computing strain energy density.
Cauchy Stress (a.k.a. True Stress)

It's been claimed that among the many
different strain definitions, no one is
necessarily superior to another. They are all
just different, each having pros and cons for a
given application. However, I don't believe that
is the case for stress. For stresses, the Cauchy
or True Stress definition appears to be head-
and-shoulders more relevant, physical,
justified, etc, over other definitions.
Cauchy stress is "force over area in the

deformed configuration". So the object has
rotated and deformed. And as it has deformed,
its cross-sectional area has changed from the
undeformed configuration. But it is this rotated
and deformed condition that is in final
equilibrium, not the reference configuration.
Cauchy stress is represented simply by . All

other stress definitions have some kind of sub
or superscripts to indicate what they are.
A very-flimsy diving board is an excellent example of a loading condition that supports this argument. The
deformed shape is in equilibrium, not the reference shape. So equilibrium equations involving stresses
should be written in terms of the deformed shape, so the Cauchy stress is the natural choice.
Work, Energy, and Power

http://www.continuummechanics.org/energeticconjugates.html 1/17
Work, Energy, and Power
It is well known that work is the dot-product of force and displacement according to
W = F dx
and power is the time-derivative of work, which can be obtained by first taking an infinitesimal amount of
work, dW = F dx, and dividing through by dt to obtain
dW dx
P = = F
dt dt
But dx/dt is simply velocity, v . So power is
P = F v
For mechanics calculations, it is often desirable to calculate work and energy in terms of stress and strain
rather than force and displacement. Likewise, calculations of power are made in terms of stress and strain-
rate instead of force and velocity. The following derivation shows how the quantities are related.
Begin by considering the power generated by forces, both external and internal, acting on an object and
moving it at velocity, v . The external forces will be expressed as traction vectors, T, acting on the outer
surface of the object. The traction vectors must be integrated over the outer surface to obtain force. The
internal forces will be represented by f , having dimensions of force/volume. They arise due to mechanisms
such as gravity, accelerations, magnatism, etc. The total force acting on the object will be
Total Force = T dA + f dV

External Internal
Forces Forces
The power generated by the forces is computed by simply performing dot-products with the velocity vector,
v.
Power = T v dA + f v dV
Note that v need not be constant over the volume (or surface). It can vary due to deformations, rotations,
and/or vibrations. Nevertheless, the equation remains correct.
This result can be partitioned into the power associated with motion (involving the velocity vector, v ), and
that associated with deformations (stresses and strains). This is accomplished in a few steps as follows.
First, use the traction vector identity, T = n , to replace T in the equation.
Power = ( n) v dA + f v dV
Then apply the divergence theorem to the first term containing the surface integral to transform it into a
volume integral.
Power = ( v) dV + f v dV
Next, expand the divergence operator to obtain
Power = ( ) v dV + : v dV + f v dV
The above step is not at all intuitive using matrix notation, but is easily verified with tensor
notation. The term, ( v) is expressed in tensor notation as (ij vi ),j . Applying the
product rule gives
(ij vi ),j = ij ,j vi + ij vi ,j
which is easily interpreted to be
( v) = ( ) v + : v
The term is key here because it also appears in the equilibrium equation: + f . This
= a
makes it possible to replace with ( a f ).
Power = ( a f ) v dV + : v dV + f v dV
Expanding the first term out gives
Power = a v dV f v dV + : v dV + f v dV
Note that the two f v dV terms cancel each other, leaving only
P dV + dV
Power = a v dV + : v dV
And the term is in fact the time derivative of Kinetic Energy: . So it can be
1
a v dV (v v) dV
2
written as
d 1
Power = ( (v v) dV ) + : v dV
dt 2
Or simply
d
Power = (Kinetic Energy) + : v dV
dt
And v is the velocity gradient, L . Substitution gives
d
Power = (Kinetic Energy) + : L dV
dt
And L is also equal to D + W. Substituting and expanding gives
d
Power = (Kinetic Energy) + : D dV + : W dV
dt
But : W is identically zero because is symmetric and W is antisymmetric. This leaves the final
result.
d
Power = (Kinetic Energy) + : D dV
dt
The total power has now been partitioned into two contributing parts: (i) bulk motion, and (ii) deformations.
The bulk motion is represented by kinetic energy, of which the velocity vector, v , is the key. Deformations
are represented by strain energy density, and its rate of change is computed by energetically conjugate
pairs of stresses and strain-rates.
We will focus on the deformation component of the total power in order to identify additional pairs of
energetically conjugate stresses, strains, and strain-rates. We already have the Cauchy stress and the rate
of deformation tensor as our first energetically conjugate pair. This is logical because both are related to
the deformed and rotated configuration of an object.
1st Piola Kirchhoff Stress

Definition of the 1st Piola Kirchhoff stress starts from the Cauchy stress and rate of deformation tensor
Definition of the 1st Piola Kirchhoff stress starts from the Cauchy stress and rate of deformation tensor.
P = : D dV
But
1
T
D = (L + L )
2
Substitute this into the power calculation.
1
T
P = : L + : L dV
2
But since is symmetric, this reduces to
P = : L dV
and L , so substitute
1
= F F
1
P = : (F F ) dV
and dV = J dVo . Substitute this to get
1
P = : (F F ) J dVo
Once again, converting everything to tensor notation helps to better understand how to regroup
components.
1
P = ij F ik F J dVo
kj
Rearrange to get
1
P = ij F F ik J dVo
kj
1 T
http://www.continuummechanics.org/energeticconjugates.html F F 5/17
and take the transpose of the transpose of Fkj to get Fjk .
1 T
T
P = J ij F
F
jk ik dVo
So this dictates that
PK1
P = : F dVo
where
PK1 T
= J F
The important result here is that the resulting P K 1 definition is NOT symmetric because, while it is
postmultiplied by FT , it is not premultiplied by a corresponding F1 to make the result symmetric. So
this is not a popular stress tensor to use.
2nd Piola Kirchhoff Stress

The derivation of the 2nd Piola Kirchhoff stress is similar to that of the 1st. Once again, start with the
Cauchy stress and the rate of deformation tensor.
P = : D dV
This time, use the relationship between D and E

.
T 1
D = F E F
Substitute this into the power calculation.
T 1
P = : (F E F ) dV
and dV = J dVo . So substitute to get
T 1
P = : (F E F ) J dVo
Once again, converting everything to tensor notation helps to better understand how to regroup
components.
T 1
P = ij F E mn F J dVo
im nj
Rearrange to get
1 T
P = J F ij F E mn dVo
mi jn
So this dictates that
PK2
P = : E dVo
where
PK2 1 T
= J F F

PK2
is symmetric and is a popular stress tensor. It is conjugate to E

for power calculations, and
conjugate to E for energy.

PK2
is in the reference configuration. Let's plug in the polar decomposition to see this more clearly.
Substitute U1 R T for F1 . Recall that R 1 = R T .
And substitute R U for F . Recall that U is symmetric, so U .

1 T T 1
= U
This produces
PK2 1 T 1
= J U R R U
This result is not necessarily useful for calculating P K 2 because it requires a polar decomposition be
performed that is otherwise not useful, but it does give insight into the stress definition. For example, the
R term rotates the Cauchy stress from the current configuration back to the reference
T
R
configuration. This will be demonstrated in examples following the section on engineering stress.
Additional Relationships
Recall the energetically conjugate pairing of the 2nd Piola-Kirchhoff stress and Green strain
tensor.
PK2
P = : E dVo
This immediately means that the following equation is also true.
PK2
W = : dE
where W is the strain energy per unit volume. (Volume is length3, hence the three
apostrophes.)
Engineering Stress
Engineering stress is simply force over initial area.
Fnormal Fparallel
Eng = and Eng =
Ao Ao
But I don't know how to extend these definitions to the case of large rotations.
Stress Examples
Examples here will compare Cauchy stress, 2nd Piola-Kirchhoff stress, and engineering stress. The
examples will use incompressible rubber, so J = 1.
Simple Tension/Compression of Incompressible Rubber
A rubber test sample is stretched in tension as shown in the figure. It is incompressible, so the
volume must remain constant.
Lo Ao = LF AF
y
Rearrange to get
Ao LF
=
AF Lo
But the ratio of initial to final lengths is

related to the engineering strain as follows
F x
LF
= 1 + Eng
Lo
So therefore
Ao
= 1 + Eng
AF
This relationship will be used to relate the different stress terms. For starters, since in this case,
the tensile force is the product of ( AF ) and also the product of (Eng Ao ) , the two terms can
be equated to give
AF = Eng Ao
And this can be rearranged to give
Ao
=
Eng AF
And this can be further manipulated to give
= Eng (1 + Eng )
So when the strains are small, the Cauchy stress and engineering stress are the same for all
practical purposes. But as an object is stretched significantly so that its cross-sectional area
decreases, the Cauchy stress will become greater than the engineering stress. Likewise, under
compression, the opposite case exists. Under compression, the Cauchy stress is less (in
absolute value) than the engineering stress.
Relating the Cauchy stress and 2nd Piola-Kirchhoff stress requires computing F and then
relating the two through
PK2 1 T
= J F F
For tension/compression of incompressible rubber, the deformation gradient is
(1 + Eng ) 0 0

1/2
F = 0 (1 + Eng ) 0

1/2
0 0 (1 + Eng )
Note that its determinant equals one because it is incompressible, J = 1 .
Its inverse is
1
(1 + Eng ) 0 0

1 1/2
F = 0 (1 + Eng ) 0

1/2
0 0 (1 + Eng )
This leads to

PK2
=
2
(1 + Eng )
and
Eng
PK2
=
1 + Eng
So the deviation of the 2nd Piola-Kirchhoff stress from the engineering stress is just the opposite
of the Cauchy stress. When the Cauchy stress is greater than the engineering stress, then P K 2
is less than Eng , and vice-versa.
The graph below assumes that Eng = E Eng with E = 1 MPa, roughly the stiffness of natural
rubber, at least at smaller strains.
Tension and Rotation of Incompressible Rubber

Recall the example from the page on true strains where the object is stretched and rotated. This
time, we will use it to compare stresses in the presence of large rigid body rotations. Engineering
stress is not included since it loses meaning in the presence of rotations.
y F
F
x F
F
Time
First, the Cauchy stress is simple. The final stress state consists entirely of a normal y-
component because the y-direction is the direction of the tensile force. So the final Cauchy
stress tensor would look like
0 0 0

= 0 F /A 0

0 0 0
where A is the final, deformed cross-section of the sample.
The deformation gradient must be determined in order to calculate the 2nd Piola-Kirchhoff
stress. The deformation gradient is most easily computed here by taking advantage of the polar
decomposition once again. This time
(1 + Eng ) 0 0


1/2
U = 0 (1 + Eng ) 0

1/2
0 0 (1 + Eng )
and the rotation matrix is
0 1 0

R = 1 0

0

0 0 1
So the deformation gradient is
0 1 0 (1 + Eng ) 0 0

R U = 1 1/2
F = 0 0 0 (1 + Eng ) 0

1/2
0 0 1 0 0 (1 + Eng )
1/2
0 (1 + Eng ) 0

= (1 + Eng ) 0 0


1/2
0 0 (1 + Eng )
The inverse of F is
1
0 (1 + Eng ) 0

1 1/2
F = (1 + Eng ) 0 0

1/2
0 0 (1 + Eng )
The 2nd Piola-Kirchhoff stress is
F /A
0 0
2
(1+ Eng )

PK2 1 T
= J F F =

0 0 0

0 0 0
The 2nd Piola-Kirchhoff stress has a nonzero component in the normal-x location because the
force acts on the object in a direction that was initially in the x-direction.
The term amounts to being /(1 + Eng )2 , but could also be written as Eng /(1 + Eng ) for
this simple tension case.

PK2
is less than while E

is greater than D by just the right amount such that PK2
: E is
exactly equal to : D. Each pair is energetically conjugate.
A Note Concerning Linear Elasticity
We've talked about Hooke's Law relating stress and strain in the section on tensor notation.
Recall that
1
= [(1 + ) I tr()]
E
But we glossed over the issue of which stress and strain tensors should be used in the equation.
The answer is deceptively simple... in most cases. This is because linear elasticity only applies
to very small strains, typically <1%. And now we've seen that all stress definitions are
equivalent, as well all strain definitions, when the strains are small. So it doesn't really matter, in
equivalent, as well all strain definitions, when the strains are small. So it doesn t really matter, in
fact.
However, there is a gotcha. It is, as usual, rotations. In the presence of large rotations, the
proper pairing of stress and strain is critical. Recall the earlier example of the object rotating
while being stretched.
y F
F
x F
F
Time
Of course, the level of stretching here is too large to be considered linear elastic. But it serves
the purpose for this discussion of rotations.
It is easy, and perfectly correct, to write the equation in terms of PK2 and E as
1
PK2 PK2
E = [(1 + ) I tr( )]
E
This will work no matter the level or rotation. Just be careful that E is strain and E is the elastic
modulus.
One could equally well write it in rate form as
1
PK2 PK2
E = [(1 + ) I tr( )]
E
Likewise, one could write Hooke's Law in terms of and True (as long as True is computed in
the current orientation).
1
True = [(1 + ) I tr()]
E
But guess what. The one thing that cannot be done (at least correctly) is to write it in terms of
and True (= D ).
To see this, consider a slight alternative of the above figure. Think of the object as first being
stretched in the x-direction, and then being held at a constant length. This produces a stress in
the x-direction that is constant while the object is held at constant length. And note that True (
= D ) is zero as a result.
So far, so good. But the problem arises when the now-stretched object starts to rotate from a
horizontal to vertical orientation. The Cauchy stress starts in the x-normal component, but
transitions to the y-normal component as the object rotates. So there are definitely non-zero
components in because the components of are changing with time. But there is no D at all
because the object is not stretching; it is only rotating.
So here we have a major problem in that is not zero while D is zero. This leads to the issue
of corotational derivatives, which address this disparity. We will cover them in more detail a little
later.

1 Login
Sort by Oldest
Recommend 1 Share
Join the discussion
LOG IN WITH
Name
hamed 9 months ago

Thank you very much Dr. McGinty for this useful page. I have a question about relating divergence of
sigma and v production to two separate terms in "Work, Energy, and Power" section ate the top of the
page. You presented divergence of dot product of stress tensor and velocity v as (ij.vi),j but I think it
must be corrected as (ijvj),i. Is it right or I am mistaking somewhere? and if the second term is correct
what happens to the remains of the equations. bests
Reply Share
BobMcG Moderator > hamed 9 months ago

Thanks for your interest Hamed. I see your point about the subscripts. If sigma were not
symmetric, then what you say would be absolutely necessary. But since it is symmetric, I took
advantage of that to swap the "i" and "j" due to personal preference. In this special case, you
would get the same answer either way because of the fact that sigma is indeed symmetric.
Reply Share
Wandy Zhang 6 months ago

Dear Bob,
This is the best tutorial I have ever leaned regarding to the Continuum Mechanics. Thank you for
providing these. I have a question in "Work, Energy, and Power" section, you wrote that: "The external
forces will be expressed as traction vectors, T, acting on the outer surface of the object. The traction
vectors must be integrated over the outer surface to obtain force." But from the previous chapter "Home
> Stress > Traction vector" you have mentioned that "The traction vector, T, is simply the force vector on
a cross-section divided by that cross-section's area." I am quite confused, why we integrate over the
outer surface area of the deformable body, not the cross-section area.
Best regards,
Wandy
Reply Share
BobMcG Moderator > Wandy Zhang 6 months ago

Wandy - Here's why you don't integrate the traction vectors over a cross-section. It's because a
cross-section cuts through an object, so it is in the interior, which is not where the external traction
vectors are acting. The traction vectors are all on the outer surface, so it is the outer surface area
that you must integrate over. Compare the traction vectors to pressure that is acting on the outer
surface of an object. You would then need to integrate the pressure over the outer surface to get
the net force. We're doing essentially the same thing here with traction vectors. Bottom line, a
traction vector is a force per unit area. It doesn't matter whether that unit area is internal like a
cross-section, or external like surface area.
Reply Share
Wandy Zhang > BobMcG 6 months ago

Thanks, here is what I understand after you enlightening me: the external traction vector
(surface-area) is related to the pressure (whatever positive or negative) acting on the
configuration's outside surface and lead to the external work; while the internal traction
vectors (cross-area) which is something related to stress and resulting to the internal work
inside the body.
Reply Share
BobMcG Moderator > Wandy Zhang 6 months ago

One correction. A traction vector is more than just pressure because, while
pressure only acts normal to a surface, a traction vector can act in any direction on
a surface.
Reply Share
Wandy Zhang > BobMcG 6 months ago

Yep, thx
Reply Share
Anup 2 months ago

Thank you for this great resource, Dr. McGinty. I had a question about the partial derivatives in the
divergence operator- How do I know if the derivatives are with respect to the reference configuration or
the current configuration i.e. d/dX or d/dx?
While it is implicit when you substitute L for the divergence of the velocity field (which is with respect to
the current configuration), how would I know in an arbitrary situation?
Reply Share
BobMcG Moderator 2 months ago

Anup - It is standard practice for derivatives for the deformation gradient, F, to be with respect to the
reference coordinates just as you noted that it is standard for derivatives for the velocity gradient, L, to be
with respect to the current configuration. And like you say, it should be written explicitly as d/dX or d/dx,
which would be clear. The main problem arises when someone writes x_i,j. In this case it is not clear
whether derivative w.r.t the j coordinate is in the ref or current configuration. In such cases, it must be
determined from context, i.e., ref config if talking about F and current config if talking about L.
Reply Share
Reply Share
hosting fees.
formats. formats.
Bob McGinty
bmcginty@gmail.com
Table of Contents
Traction Vectors Stress Transformations
12/20/2017 Stress Transformations

Stress Transformations Search
home > stress > stress transformations
Introduction
As with strain, transformations of stress tensors follow the same rules of pre and post multiplying by a
transformation or rotation matrix regardless of which stress or strain definition one is using. The only
difference is that the full shear values, ij , are used in stress tensors and their transformations, not the half
shear values, /2, used in strain tensors.
This page will cover coordinate transformations and rotations in 2-D and 3-D.
2-D Coordinate Transforms of Stress

Coordinate transforms represent rotations of the
coordinate system while the object is held
constant. The common application of coordinate A cosO A
transforms is to rotate the coordinate system to
find the principal directions of the stress tensor. O
The governing equations are derived by
summing forces on differential objects. The
sxx t' s'
sketch here demonstrates this for the (relatively)
simple 2-D case. y
Summing forces in the x and y directions gives

x
txy
syy
A sinO

xx A cos + xy A sin = A cos A sin

xy A cos + yy A sin = A sin + A cos
The area, A, cancels out of both sides leaving

xx cos + xy sin = cos sin
http://www.continuummechanics.org/stressxforms.html 1/6

xy cos + yy sin = sin + cos
This represents two simultaneous equations with two unknowns, and . Solving the equations gives
the well-known 2-D stress transformation equations.
2 2
= xx cos + yy sin + 2 xy sin cos
2 2
= (yy xx ) sin cos + xy (cos sin )
This gives the normal and shear stresses on any face and is, in fact, the exact same result as obtained on
the earlier page on traction vectors, except this is 2-D, of course. The complete transformed 2-D stress
state is obtained by evaluating these two equations, plus one additional equation for normal stress at
+ 90 .

2 2
xx = xx cos + yy sin + 2 xy sin cos
2 2
yy = xx sin + yy cos 2 xy sin cos
2 2
xy = (yy xx ) sin cos + xy (cos sin )
The complete set of transformation equations can be summarized in matrix form as

xx xy cos sin xx xy cos sin
[ ] = [ ][ ][ ]

xy yy sin cos xy yy sin cos
The leading and trailing matrices are the familiar Q matrix and its transpose. The complete set of
equations can be written simply as
T
= Q Q
2-D Stress Transform Example
1 2
If the stress tensor in a reference coordinate system is [ ] , then in a coordinate system
2 3
rotated 50, it would be written as

xx xy cos 50 sin 50 1 2 cos 50 sin 50
[ ] = [ ][ ][ ]

xy yy sin 50 cos 50 2 3 sin 50 cos 50
4.143 0.638
= [ ]
0.638 0.143
The stress state has not changed at all. Only the values of the individual components are
different because the orientations of the coordinate systems are different.
Tensor Notation
The coordinate transform is written in tensor notation as

mn = mi nj ij
As usual, tensor notation provides extra insight into the process. This time, the insight comes from the
subscripts on the lambdas. Each lambda effectively pairs up a subscript on with one on . This is true
regardless of the rank of the tensor.
3-D Coordinate Transforms

Once again, the rules don't change, only the particulars do.
T
= Q Q mn = mi nj ij ij = cos(x , xj )
i
Writing the matrices out explicitly gives

Q11 Q12 Q13 xx xy xz
xx xy xz
Q11 Q21 Q31

xy yy yz = Q21 Q22 Q23 xy yy yz Q12 Q22 Q32

xz yz zz
Q31 Q32 Q33 xz yz zz Q13 Q23 Q33
This webpage performs coordinate transforms on 3-D tensors. Try it out.
Rotations
Rotations are the process in which the object rotates while the coordinate system remains fixed. The
rotation matrix, R , is usually computed from a polar decomposition. The rotated stress tensor is calculated
as
T
= R R
which looks like this in 3-D.

R 11 R 12 R 13 xx xy xz
xx xy xz
R 11 R 21 R 31

xy yy yz = R 21 R 22 R 23 xy yy yz R 12 R 22 R 32

xz yz zz R 31 R 32 R 33 xz yz zz R 13 R 23 R 33
In 2-D, the rotation matrix is the transpose of the coordinate transformation matrix.
cos sin cos sin

R = [ ] Q = [ ]
sin cos sin cos
So for a 2-D rigid body rotation of degrees, the equations are
2 2
xx
= xx cos + yy sin 2 xy sin cos
2 2
yy = xx sin + yy cos + 2 xy sin cos
2 2
xy
= (xx yy ) sin cos + xy (cos sin )
xy
(xx yy ) s cos + xy (cos sin )
2-D Stress Rotation Example

Take the coordinate transformation example from above and this time apply a rigid body rotation
of 50 instead of a coordinate transformation.
1 2
If the stress tensor in a reference coordinate system is [ , then after rotating 50, it would
]
2 3
be

xx xy cos 50 sin 50 1 2 cos 50 sin 50
[ ] = [ ][ ][ ]

xy yy sin 50 cos 50 2 3 sin 50 cos 50
0.204 1.332
= [ ]
1.332 3.796
The stress state has not changed at all. Only the values of the individual components are
different because the object has rotated relative to the global coordinate system.
Rotating and Transforming Back

Start with the above rotation equation.
T
= R R
Premultiply both sides by R and postmultiply both sides by R .

T T T
R R = R R R R
The R and R T matrices cancel out (because R T ) leaving

1
= R
T
R R =
The process of transforming a coordinate system back to its original orientation is exactly the same. The
equation to do this is
T
= Q Q

1 Login
Sort by Oldest
Recommend 3 Share
LOG IN WITH
Name
hosting fees.
formats. formats.
Bob McGinty
bmcginty@gmail.com
Table of Contents
Energetic Conjugates Principal Stresses
12/20/2017 Principal Stresses

Principal Stresses & Search
Invariants
home > stress > principal stress
Introduction
This page covers principal stresses and stress invariants. Everything here applies regardless of the type of
stress tensor.
Coordinate transformations of 2nd rank tensors were discussed on this coordinate transform page. The
transform applies to any stress tensor, or strain tensor for that matter. It is written as
T
= Q Q
Everything below follows from two facts: First, the tensors are symmetric. Second, the above coordinate
transformation is used.
2-D Principal Stresses

In 2-D, the transformation equations are
2 2
xx = xx cos + yy sin + 2 xy sin cos
2 2
yy = xx sin + yy cos 2 xy sin cos
2 2
The full shear stress values are used, unlike strain transformations, which use half values for shear strain,
i.e., (/2).
This page performs full 3-D tensor transforms, but can still be used for 2-D problems.. Enter values in the
upper left 2x2 positions and rotate in the 1-2 plane to perform transforms in 2-D. The screenshot below
shows a case of pure shear rotated 45 to obtain the principal stresses. Note also how the Q matrix
transforms.
http://www.continuummechanics.org/principalstress.html 1/15
The figure below shows the stresses corresponding to the pure shear case in the tensor transform
webpage example. The blue square aligned with the axes clearly undergoes shear. But the red square
inscribed in the larger blue square only sees simple tension and compression. These are the principal
values of the pure shear case in the global coordinate system.
tyx
s2 s1
txy txy
y
s1 s2
x tyx
In 2-D, the principal stress orientation, P , can be computed by setting xy = 0 in the above shear
equation and solving for to get P , the principal stress angle.
2 2
0 = (yy xx ) sin P cos P + xy (cos P sin P )
Thi i
This gives
2xy
tan(2P ) =
xx yy
The transformation matrix, Q , is
cos P sin P
Q = [ ]
sin P cos P
Inserting this value for P back into the equations for the normal stresses gives the principal values. They
are written as max and min , or alternatively as 1 and 2 .

2
xx + yy xx yy
2
max , min = ( ) + xy
2 2
They could also be obtained by using = Q Q

T
with Q based on P .
Principal Stress Notation
Principal stresses can be written as 1 , 2 , and 3 . Only one subscript is usually used in this
case to differentiate the principal stress values from the normal stress components: 11 , 22 ,
and 33 .
2-D Principal Stress Example

Start with the stress tensor
50 30
= [ ]
30 20
The principal orientation is
2 30
tan(2P ) =
50 (-20)
syy = -20
txy = 30
xy
P = 20.3
y
The principal stresses are
x sxx = 50

2
50 20 50 + 20
2
max , min = ( ) + (30)
2 2
max , min = 61.1, 31.1
The only limitation to using these equations for principal values is that it is not known which one
applies to 20.3 and which applies to 110.3. That's why an attractive alternative is

11 12
cos(20.3 ) sin(20.3 ) 50 30 cos(20.3 ) sin(20.3 )
[ ] = [ ][ ][ ]

12 22 sin(20.3 ) cos(20.3 ) 30 20 sin(20.3 ) cos(20.3 )
61.1 0.0
= [ ]
0.0 31.1
This confirms that the 61.1 principal stress value in the 11 slot is indeed 20.3 from the X-axis.
The 22 value is 90 from the first.
3-D Principal Stresses

Coordinate transforms in 3-D are

q 11 q 12 q 13 11 12 13 q 11 q 21 q 31
11 12 13

=
12 q 21 q 22 q 23 12 22 23 q 12 q 22 q 32
22 23

q 31 q 32 q 33 13 23 33 q 13 q 23 q 33
13 23 33
The second Q matrix is once again the transpose of the first

The second Q matrix is once again the transpose of the first.
This page performs tensor transforms.
And this page calculates principal values (eigenvalues) and principal directions (eigenvectors).
It's important to remember that the inputs to both pages must be symmetric. In fact, both pages enforce
this.
The eigenvalues above can be written in matrix form as
24 0 0

= 0 125 0

0 0 433
Maximum Shear Stress
The maximum shear stress at any point is easy to calculate from the principal stresses. It is
simply
max min
max =
2
This applies in both 2-D and 3-D. The maximum shear always occurs in a coordinate system
orientation that is rotated 45 from the principal coordinate system. For the principal stress
tensor above
24 0 0

= 0 125 0

0 0 433
The max and min principal stresses are in the 33 and 11 slots, respectively. So the max shear
orientation is obtained by rotating the principal coordinate system by 45 in the (1 3) plane.
The max shear value itself is
max min
max =
2
= (433 24)/2
= 204
The manual way of computing principal stresses is to solve a cubic equation for the three principal values.
The equation results from setting the following determinant equal to zero. The values, once computed,
will equal the principal values of the stress tensor.
12 13
11

12 22 23 = 0

13 23 33
This can be expanded out to give
2
(11 )[(22 )(33 ) ]
23
12 [12 (33 ) 23 13 ] +
13 [12 23 (22 )13 ] = 0
Invariants
...and expanded out even further to give
3 2 2 2 2
(11 + 22 + 33 ) + (11 22 + 22 33 + 33 11 )
12 13 23
2 2 2
(11 22 33 11 22 33 + 212 13 23 ) = 0
23 13 12
"Now consider this..."
No matter what coordinate transformation you apply to the stress tensor, its principal stress had
better be the same three values. And the only way for this to happen in the above equation is for
the equation itself to always be the same, no matter the transformation. This means that the
combinations of stress components, which serve as coefficients of the 's, must be invariant
under coordinate transformations. Their values must not change. And that's why they are called
invariants.
So the equation can be written as
3 2
I1 + I2 I3 = 0
where
where
I1 = 11 + 22 + 33
2 2 2
I2 = 11 22 + 22 33 + 33 11
12 13 23
2 2 2
I3 = 11 22 33 11 22 33 + 212 13 23
23 13 12
Although probably not initially apparent, the invariants are actually familiar quantities.
I1 = tr()
12 13 23
11 11 22
I2 = + +
12 22 13 33 23 33
I3 = det()
They can be written in tensor notation as
I1 = kk
1 2
I2 = [(kk ) ij ij ]
2
I3 = ijk i1 j2 k3
I3 is in tensor notation, but no one should actually calculate a determinant based on tensor notation rules
because it is very inefficient.
Physical Interpretation of Invariants

The physical interpretation of the invariants depends on what tensor the invariants are computed
from. For any stress or strain tensor, I1 is directly related to the hydrostatic component of that
tensor. This is universal.
I2 tends to be related more to the deviatoric aspects of stress and strain. For stress tensors, it is
closely related to the von Mises stress.
Finally, I3 does not seem to have any physical significance as the determinant of a stress or
strain tensor. But it does when applied to the deformation gradient. In that case, I3 = VF /Vo ,
the ratio of deformed to initial volume, which is 1 for incompressible materials like rubber.
Invariant Example
This example will start with a random stress tensor and demonstrate that the invariants are
indeed invariant under coordinate transformations. Start with the stress tensor
50 30 20

= 30 20 10

20 10 10
Its invariants are
I1 = 50 + (20) + 10
= 40
2 2 2
I2 = (50)(20) + (20)(10) + (10)(50) (30) (20) (10)
= 2, 100
I3 = det()
= 28, 000
Now rotate the coordinate system by some random amount shown in the screenshot.
The transformed stress tensor is now
30.8 10.6 27.3

= 10.6 34.6 03.5

27.3 3.5 43.7
And the "new" invariants are
I1 = 30.8 + (34.6) + 43.7
= 40
2 2 2
I2 = (30.8)(34.6) + (34.6)(43.7) + (43.7)(30.8) (10.6) (27.3) (3.5)
= 2, 100
I3 = det()
= 28, 000
So they are indeed invariant.
Just for grins... Let's compute the principal stresses and then recompute the invariants to
demonstrate again that they are invariant.
The principal stresses are
65.5 0 0

= 0 37.1 0

0 0 11.5
Curiously, the two different starting points lead to the principal values being listed in different
orders. This is typical and has no special meaning. The final, physical answer is the same. The
Q will accommodate the different listing orders.
Finally, calculating the invariants one last time using the principal values gives
I1 = 65.5 + (37.1) + 11.5
= 40
I2 = (65.5)(37.1) + (37.1)(11.5) + (11.5)(65.5)
= 2, 100

I3 = det( ) = (65.5)(37.1)(11.5)
= 28, 000
And so it works again. Note how the invariant calculations are rather simple when all the off-
diagonal terms are zero. The determinant is especially easy.
g p y y
2-D Invariants
In the 2-D world, there are only two invariants.
I1 = tr() = 11 + 22
2
I2 = det() = 11 22
12
In 2-D, things are just simple enough to directly prove the invariance of invariants. We'll prove it
for I1 here. Recall the transformation equations are
2 2
xx = xx cos + yy sin + 2xy sin cos
2 2
yy = xx sin + yy cos 2xy sin cos
2 2
So xx +

yy is
2 2
xx + yy = xx cos + yy sin + 2xy sin cos +
2 2
xx sin + yy cos 2xy sin cos
and the shear terms cancel, while the normal stress terms combine to give
2 2 2 2
xx
+ yy
= xx (cos + sin ) + yy (sin + cos )
= xx + yy
So it works again.
Calculating Principal Values Manually in 3-D

Here's how to compute the roots of a cubic equation... applied to principal values. Start with
3 2
I + I I 0
3 2
I1 + I2 I3 = 0
Recall that the resulting s will be the principal values of the stress or strain tensor... and the
invariants will need to be calculated up front.
The first step is to calculate two intermediate quantities, Q and R.
2 3
3I2 I 2I 9I1 I2 + 27I3
1 1
Q = R =
9 54
Then calculate another intermediate quantity, .
R
1
= cos ( )

3
Q
And finally, the principal values are
1
1 = 2 Q cos( ) + I1
3 3
+ 2 1
2 = 2 Q cos( ) + I1
3 3
+ 4 1
3 = 2 Q cos( ) + I1
3 3
Note that minus signs are easily confused here because they are present in the original cubic
invariant equation being solved.
Ref: http://mathworld.wolfram.com/CubicFormula.html
Summary
This principle of invariant quantities under
coordinate transformations is in fact universal across all matrices that are symmetric and being
transformed according to
T
A = Q A Q
where A is "any symmetric matrix."
And recall that the product of any matrix with its transpose is always a symmetric result, so this result
would qualify. This is especially relevant to FT F , whose invariants are used in the Mooney-Rivlin Law of
rubber behavior. Mooney-Rivlin's Law and coefficients will be discussed on this page. As an added teaser,
we will see that the 3rd invariant of F F for rubber always equals 1 because rubber is incompressible.
T
So not only is it a constant, independent of coordinate transformations, but it is even a constant value,
always equal to 1, independent of coordinate transformations and the state of deformation.

1 Login
Sort by Oldest
Recommend Share
LOG IN WITH
Name
hosting fees.
formats. formats.
Bob McGinty
bmcginty@gmail.com
Table of Contents
Stress Transforms Hydro & Deviatoric Stresses
12/20/2017 Hydrostatic & Deviatoric Stresses

Hydrostatic & Deviatoric Search
Stresses
home > stress > hydrostatic stress
Introduction
This page introduces hydrostatic and deviatoric stresses. The two are subsets of any given stress tensor,
which, when added together, give the original stress tensor back. The hydrostatic stress is related to
volume change, while the deviatoric stress is related to shape change.
Hydrostatic Stress
y
Hydrostatic stress is simply the average of the three normal
stress components of any stress tensor. s Hyd
11 + 22 + 33
Hyd =
3
s Hyd s Hyd
There are many alternative ways to write this.
1 1 1 s Hyd
Hyd = tr() = I1 = kk
3 3 3
x
It is a scalar quantity, although it is regularly used in tensor
form as
Hyd 0 0

Hyd = 0 Hyd 0

0 0 Hyd
Hydrostatic Stress Example
Given the following stress tensor
http://www.continuummechanics.org/hydrodeviatoricstress.html 1/6
50 30 20

= 30 20 10

20 10 10
The hydrostatic stress is
50 + (20) + 10
Hyd = = 13.3
3
which can be written as
13.3 0 0

Hyd = 0 13.3 0

0 0 13.3
And that's all there is to it.
Hydrostatic Stresses and Coordinate Transformations

This could not be simpler. Hydrostatic stresses do not change under coordinate transformations. This is
easily accepted in light of the fact that Hyd is a function of I1 . Also
Hyd 0 0

Hyd = 0 Hyd 0

0 0 Hyd
contains equal amounts of stress in all three directions. And since the tensor does not change under any
transformation, this means that no shear stresses ever arise, so every direction is a principal direction with
Hyd stress.
Hydrostatic Stress and Pressure

Pressure is simply the negative of hydrostatic stress. The negative aspect is often confusing. It is why we
talk about atmospheric pressure as 30 inches of Hg, a positive number, even though atmospheric pressure
is in fact a negative stress because it is compressive. So using P for pressure...
(11 + 22 + 33 )
(11 + 22 + 33 )
P = Hyd =
3
The stress tensor containing pressure, P , is
P 0 0

Hyd = 0 P 0

0 0 P
Of course, it is rare to talk about pressure unless the hydrostatic stress is compressive, which corresponds
to a positive pressure.
Also, unless one is working with boundary layer flows over aircraft, automobiles, etc, then the stress state
in the air is one of hydrostatic stress alone, without any shear stresses. And the hydrostatic stress is
compressive, which is a positive pressure.
Hydrostatic Stress in Incompressible Materials

This is almost a nontopic because hydrostatic stresses usually have no impact on incompressible materials
at all, so there is very little to discuss. This also means that one cannot determine the hydrostatic stress
based on strain or a deformation gradient. And this is why finite element programs often fudge by making
the incompressible material ever-so-slightly compressible.
Deviatoric Stress
Deviatoric stress is what's left after subtracting out the hydrostatic stress. The deviatoric stress will be
represented by . For example

= Hyd
In tensor notation, it is written as
1

ij = ij ij kk
3
And in terms of pressure, it is written as

ij = ij + P ij
Deviatoric Stress Example

Given the following stress tensor
50 30 20

= 30 20 10

20 10 10
The hydrostatic stress is
50 + (20) + 10
Hyd = = 13.3
3
which can be written as
13.3 0 0

Hyd = 0 13.3 0

0 0 13.3
Subtracting the hydrostatic stress tensor from the total stress gives
50 30 20 13.3 0 0 36.7 30.0 20.0

= 30 20 10 0 13.3 0 = 30.0 33.3 10.0

20 10 10 0 0 13.3 20.0 10.0 3.3
Note that the result is traceless. Its first invariant equals zero. Or put another way, the
hydrostatic stress of a deviatoric stress tensor is zero.
An interesting aspect of a traceless tensor is that it can be formed entirely from shear components. For
example, a coordinate system transformation can be found to express the deviatoric stress tensor in the
above example as shear stress exclusively. In the screenshot here, the above deviatoric stress tensor was
input into the webpage, and then the coordinate system was rotated until the following stress tensor was
obtained.

1 Login
Sort by Oldest
Recommend 4 Share
Join the discussion
LOG IN WITH
Name
Rajesh Prasad 14 days ago

How to find the coordinate transformation which will transform a traceless deviatoric stress tensor to one
having shear components only?
Reply Share
BobMcG Moderator 14 days ago

I think the only way to do that is iteratetively. You can try to iterate with this webpage:
http://www.continuummechani...
On the other hand, going backwards (from shears to traceless principals) can be done by first entering
the tensor containing only shear strains into this form:
http://www.continuummechani...
to compute principal values and the eigenvectors, which combine to form the coordinate xform matrix.
Reply Share
ebpage ee ee to e a ( (
hosting fees.
formats. formats.
Bob McGinty
bmcginty@gmail.com
Table of Contents
Principal Stresses Von Mises Stress
12/20/2017 Von Mises Stress

Von Mises Stress Search
home > stress > von mises stress
Introduction
Von Mises
Yield Surface
The von Mises stress is often used in determining
whether an isotropic and ductile metal will yield Hydrostatic
Axis
when subjected to a complex loading condition.
This is accomplished by calculating the von Mises
stress and comparing it to the material's yield Von Mises
Yield Curve
stress, which constitutes the von Mises Yield
Criterion.
The objective is to develop a yield criterion for

ductile metals that works for any complex 3-D
loading condition, regardless of the mix of plane
(Deviatoric Plane)
normal and shear stresses. The von Mises stress
does this by boiling the complex stress state down
into a single scalar number that is compared to a
metal's yield strength, also a single scalar
numerical value determined from a uniaxial tension ref: http://en.wikipedia.org/wiki/File:Yield_surfaces.svg
test (because that's the easiest) on the material in
a lab.
It should be noted that this is not an exact science like, say F = m a. It is an empirical process, with
inherent error and deviations. In fact, there is no hard & fast rule saying that metals must yield according to
von Mises yield criteria. It is as much a coincidence as anything. Nevertheless, it does work very well and
remains the method of choice a full century after it was first proposed.
History
The defining equation for the von Mises stress was first proposed by Huber [1] in 1904, but apparently
received little attention until von Mises [2] proposed it again in 1913. However, Huber and von Mises'
definition was little more than a math equation without physical interpretation until 1924 when Hencky [3]
recognized that it is actually related to deviatoric strain energy.
In 1931, Taylor and Quinney [4] published results of tests on copper, aluminum, and mild steel
demonstrating that the von Mises stress is a more accurate predictor of the onset of metal yielding than the
maximum shear stress criterion, which had been proposed by Tresca [5] in 1864 and was the best
predictor of metal yielding to date. Today, the von Mises stress is sometimes referred to as the Huber-
Mises stress in recognition of Huber's contribution to its development. It is also called Mises effective
stress and simply effective stress.
http://www.continuummechanics.org/vonmisesstress.html 1/16
Technical Background
A complete understanding of the von Mises stress requires an understanding of hydrostatic and deviatoric
components of stress and strain tensors, Hooke's Law, and strain energy density. The hydrostatic and
deviatoric stresses and strains have already been reviewed. And Hooke's Law has already been touched
on here and here, but will need to be discussed in additional detail on this page as well.
Strain energy density will also be introduced here.
Hydrostatic and Deviatoric Components

Recall that any stress tensor can be decomposed into the sum of hydrostatic and deviatoric stresses as
follows
1

ij = ij kk + ij
3
where 1
3
ij kk is the hydrostatic term and is the deviatoric stress.
The same is true for strain.
1

ij = ij kk + ij
3
where is the hydrostatic term and is the deviatoric strain.

1
ij kk
3
These two will be multiplied together farther down the page.
Hooke's Law
We've seen that Hooke's Law can be written as
1
ij = [(1 + )ij ij kk ]
E
which is shorthand for
xx xy xz xx xy xz hyd 0 0

1
xy yy yz = (1 + ) xy yy yz 3 0 hyd 0
E

http://www.continuummechanics.org/vonmisesstress.html 0 0 2/16

xz yz zz xz yz zz 0 0 hyd
which is in turn matrix notation for the following set of equations
1
xx = [xx (yy + zz )]
E
1
yy = [yy (xx + zz )]
E
1
zz = [zz (xx + yy )]
E
for the normal terms, and
1 + 1 + 1 +
xy = xy yz = yz xz = xz
E E E
for the shear terms. The shear terms are more commonly written as
xy yz xz
xy = yz = xz =
G G G
where
E
xy = 2xy yz = 2yz xz = 2xz and G =
2(1 + )
Return now to Hooke's Law in tensor form
1
ij = [(1 + )ij ij kk ]
E
and multiply both sides by ij .
1
ij ij = [(1 + )ij ij kk ] ij
E
This simplifies to
(1 2 )
kk = kk
E
And then multiply both sides by again to get

1
ij
3
1 (1 2 )
ij kk = ij kk
3 3E
This results in an equation relating the hydrostatic stress and strain values.
Now subtract the above equation from the original Hooke's Law equation to get
1 (1 + ) (1 2 )
ij ij kk = ij ij kk ij kk
3 E E 3E
(1 + ) 1 1 2
= ij ( + ) ij kk
E E 3
(1 + ) 1 (1 + )
= ij ij kk
E 3 E
(1 + ) 1
= (ij ij kk )
E 3
The remarkable result is that both sides of the equation contain a deviatoric tensor result. The equation
can be summarized as
(1 + )

ij = ij
E
(1+)
But is , so the equation can be further simplified to
1
E 2G
1

ij = ij
2G
So the deviatoric stress and strain are directly proportional to each other. The amazing thing here is that
this is always true for Hooke's Law, always, even for the normal strain components.
For what it's worth, the equation can also be written as

ij = 2 G ij
Deviatoric Example with Hooke's Law
Suppose you have a material with Poisson's ratio, = 0.5 , and elastic modulus, E = 15 M P a
.
For the stress tensor below, use Hooke's Law to calculate the strain state. Then get the
deviatoric stress and strain tensors and show that they are proportional to each other by the
factor 2G.
8 2 4

= 2 6 6

4 6 4
Note that this stress tensor clearly has a significant amount of hydrostatic stress. It is
6 0 0

Hyd = 0 6 0

0 0 6
Hooke's Law is
xx xy xz 8 2 4 6 0 0

1
xy yy yz = (1 + 0.5) 2 6 6 3 (0.5) 0 6 0
15

xz yz zz 4 6 4 0 0 6
12 3 6 9 0 0

1
= 3 9 9 0 9 0
15

6 9 6 0 0 9
0.2 0.2 0.4

= 0.2 0.0 0.6

0.4 0.6 -0.2
Note that this strain tensor is already deviatoric. This is because we used = 0.5 for the
Poisson's Ratio, which is the value used for incompressible materials. So we obtained an
incompressible, non-hydrostatic strain tensor as a result.
So the question becomes, "Will (2 G ) give ?"
To answer this, first compute G.
E 15MPa
G = =
2(1 + ) 2(1 + 0.5)
= 5MPa
So 2 G equals
2 2 4

2G = 2 0 6

4 6 -2
And compare this to Hyd
8 2 4 6 0 0 2 2 4

Hyd = 2 6 6 0 6 0 = 2 0 6

4 6 4 0 0 6 4 6 -2

http://www.continuummechanics.org/vonmisesstress.html2 G 6/16
So it does indeed satisfy = 2 G . Although this was an example with an incompressible

material, = 0.5, it also works for compressible materials as well, always.
Strain Energy Density

Strain energy density, W, has units of Energy / Volume and is
W = : d
For linear elastic materials, this equals
1
W = :
2
which expands out to give
1 1
: = [xx xx + yy yy + zz zz + 2(xy xy + yz yz + xz xz )]
2 2
But since = Hyd +

and = Hyd +

, these identities can be substituted into the equation to
obtain
1 1

W = : = (Hyd + ) : (Hyd + )
2 2
and expanding the multiplication out gives
1 1 1 1 1

W = : = Hyd : Hyd + Hyd : + : Hyd + :
2 2 2 2 2
But (Hyd : ) and ( : Hyd ) are zero! This is because the double-dot product of any hydrostatic tensor
with a deviatoric tensor is always zero. So the equation reduces to
1 1 1

W = : = Hyd : Hyd + :
2 2 2

hydrostatic deviatoric
This shows that strain energy can be partitioned into hydrostatic and deviatoric components.
Von Mises Stress

The von Mises stress is directly related to the deviatoric strain energy term in the above equation.
1

W = :
2
Recall from the section on Hooke's Law that
1

=
2G
Combining the two gives
1

W = :
4G
So the deviatoric part of the strain energy density is directly related to the double dot product of the
deviatoric stress with itself. Note the similarity to Kinetic Energy, K E = 2 M v2 , a spring's internal energy,
1
, electrical power, P , and any other form one can think of.
1 2 2
E = Kx = RI
2
It is finally time to introduce an equivalent or effective stress that will turn out to be proportional to the von
Mises stress, though about 20% low. Use the symbol Rep for representative stress to represent this
stress value. And it is a scalar stress value, not a tensor! The defining equation for Rep is
1
2
W = (Rep )
4G
The form of the equation is deliberately chosen to be the scalar equivalent of the one above. Setting them
equal to each other (since both are equal to W') gives
1 1
2
W = (Rep ) = :
4G 4G
Clearly Rep is intended to be the scalar stress value that gives the same deviatoric strain energy as the
actual 3-D stress tensor. Cancelling 4 G from both sides gives

Rep = :
The final step is one of simple convenience. It is motivated by the simplest straight-forward case of uniaxial
tension. To see it, calculate Rep for this case. The stress state for uniaxial tension is
0 0

= 0 0 0

0 0 0
The hydrostatic stress is , and the deviatoric stress tensor is

1

3
2
0 0
3

= 0 0

3

0 0
3
So
: equals 2 2 /3. And therefore

2

Rep = : =
3
And therein lies the frustration. The representative stress for uniaxial tension is not equal to the uniaxial
tension stress, , but is instead about 82% of it. This is terribly inconvenient, but the fix is simple. Simply
scale the representative stress up until it equals the uniaxial tension stress. This is done by simply

multiplying Ref by 3/2 .

This is acceptable because anything proportional to : will still reflect the relationship to deviatoric
strain energy. It will just be scaled up some. The final result is the von Mises stress.

3

VM = :
2
And this is the defining equation for it.
Alternate Forms
Algebraic manipulation of the above equation gives many other equivalent forms. They are summarized
here.

1 2 2 2 2 2 2
VM = [(xx yy ) + (yy zz ) + (zz xx ) ] + 3 (xy + yz + zx )
2

2 2 2 2 2 2
VM = xx + yy + zz xx yy yy zz zz xx + 3 (xy + yz + zx )

3 1 3
2
VM = ij ij (kk ) VM = ij ij
2 2 2
In 2-D applications, zz = xz = yz = 0 . This leaves

2 2 2
VM = xx + yy xx yy + 3 xy
Tensor Manipulation of Von Mises Equation

One can (relatively) easily obtain other equations for von Mises stress thru tensor manipulations
of the equation based on deviatoric values. Starting with

3

VM = ij ij
2
and expressing ij in terms of the full stress tensor as
1

ij = ij ij kk
3
gives the following form.

3 1 1
VM = (ij ij kk ) (ij ij kk )
2 3 3
Multiplying this out gives

3 2 1
2
VM = (ij ij ij ij kk + ij ij (kk ) )
2 3 9
which simplifies down to

3 1
2
VM = ij ij (kk )
2 2
The other forms listed above can be obtained by expressing this explicitly in terms of xx xy
The other forms listed above can be obtained by expressing this explicitly in terms of xx , xy ,
xz , etc.
Specific Loading Cases

We've already seen during the derivation above that for uniaxial tension, the von Mises stress equals the
uniaxial tension stress. But this is also (almost) true for compression as well. The only issue is that for
compression, the numerical value of the compressive stress will be negative, but the von Mises stress is
always positive because it is a square-root of a sum of stress values squared. So when one is reading a
von Mises stress of say, 10 MPa, it is impossible to know from this alone if the object is undergoing tension
or compression. One can look at the principal stress values to determine this.
Actually, some FEA post-processors will make color stress contours of a quantity call signed von Mises
stress. This has the same absolute value as the conventional von Mises stress, but the +/- sign is
determined by checking the sign of the hydrostatic stress. If it is negative, then the signed von Mises stress
is also negative.
The case of pure shear stress is most interesting. One can see from the equations above that for a pure
shear stress, xy , the von Mises stress is

VM = 3
So if a metal yields in uniaxial tension (or compression) at = 500MPa, then it will also yield in shear
at a stress that is only 58% of this, or = 290MPa.
Graphical Representations
Here again is the sketch at the top of the page. It shows a bounding surface in a 3-D principal stress
coordinate system where the von Mises stress is a constant value. (This is the so called High-Westerguard
Space.) It is based on the fact that any stress state can be converted into its principal values and
compared to this sketch. If the resulting principal stress point in the coordinate system is within the
cylinder, then the material has not yielded. If it is on the surface, then the material has yielded. And if it is
outside the cylinder, it means that you did an elastic analysis of a situation that cannot in fact be correct
because yielding would have long since taken place.
Von Mises
Yield Surface
Hydrostatic
Axis
Von Mises
Yield Curve
plane
(Deviatoric Plane)
ref: http://en.wikipedia.org/wiki/File:Yield_surfaces.svg
The remarkable result is that if you look down the 1 = 2 = 3 axis, the cross-section of the cylinder is a
perfect circle. Note that the hydrostatic stress in this situation does not show up at all.
Experimental Data
The figure here presents experimental data confirming that ductile metals yield much more consistently at
prescribed von Mises stress levels regardless of the the loading state than at any other criteria.
The graph represents a slice through the 1 2 plane with 3 = 0. Since the cylinder is cut at an angle,
it appears to be an ellipse in this situation. It is in fact still a circle. We are just looking at it at an angle.
Recall that the shear stress criterion was first proposed by Tresca in 1864, and this act is considered to
represent the birth of the field of metal plasticity research.
The one exception here is the cast iron metal. It yields, fractures in fact, at a constant maximum principal
stress criterion. This signifies that the iron is brittle and behaves more like glass than a ductile metal.
Reference: Dowling, N.E., Mechanical Behavior of Materials, Prentice Hall, 1993.
Note that the correlation here is not perfect. This is a consequence of the fact that the so-called von Mises
Yield Criterion is NOT a law of nature. It is more of a convenient coincidence. It is a consequence of
dislocation movement on millions and billions of planes of atoms sliding over each other at the atomic
scale. Those planes of atoms are all randomly oriented, and the resulting response at the macroscale is....
the von Mises yield criterion.
Contrasting Stress and Strain
We've seen how the von Mises stress is "the stress" when worrying about metal yielding and
plasticity. Recall that it is

3

VM = :
2
The next question is, "Is there a strain analog to the von Mises stress?" The answer is yes. It is
the effective strain, or sometimes the Mises effective strain. It is

2

eff = :
3
Note that it is 2/3, not 3/2. This arises because the strain tensor for uniaxial tension of an
incompressible material (which includes the plastic part of the total deformation of a metal) is
0 0

True =

0 0

2

0 0
2
and : in this case gives 3 2 /2. So it is necessary to multiply by 2/3 in order to make eff
equal to the uniaxial tension strain.
This makes it possible to more fairly compare the stress and strain states of two different
deformation modes, say tension versus shear. In fact, in a perfectly isotropic metal, plots of
effective stress versus effective strain will be indistinguishable in the plastic region regardless of
the deformation mode. Although in reality, metals usually become increasingly anisotropic after
yielding.
References
1. Huber, M.T. (1904) Czasopismo Techniczne, Lemberg, Austria, Vol. 22, pp. 181.
2. Von Mises, R. (1913) "Mechanik der Festen Korper im Plastisch Deformablen Zustand," Nachr. Ges.
Wiss. Gottingen, pp. 582.
3. Hencky, H.Z. (1924) "Zur Theorie Plasticher Deformationen und der Hierdurch im Material
Hervorgerufenen Nachspannungen," Z. Angerw. Math. Mech., Vol. 4, pp. 323.
4. Taylor, G.I., Quinney, H. (1931) "The Plastic Distortion of Metals," Phil. Trans. R. Soc., London, Vol.
A230, pp. 323.
5. Tresca, H. (1864) "Sur l'Ecoulement des Corps Solides Soumis a de Fortes Pressions," C. R. Acad.
Sci., Paris, Vol. 59, pp. 754.
6. Dowling, N.E. (1993) Mechanical Behavior of Materials, Prentice Hall.

1 Login
Sort by Oldest
Recommend 2 Share
LOG IN WITH
Name
hosting fees.
formats. formats.
Bob McGinty
bmcginty@gmail.com
Table of Contents
Hydro & Deviatoric Stresses Corotational Derivatives
12/20/2017 Corotational Derivatives

Corotational Derivatives Search
home > stress > corotational derivative
Introduction
We talked at the bottom of the energetic conjugates page about how certain pairs of F
stress and strain tensors can be used together in Hooke's Law, but that and D
should not be used together in the rate form.
This is because the Cauchy stress and rate of deformation tensor behave
incompatibly in the presence of rigid body rotations like that shown at the right. It is
probably easiest to understand the situation where the object has been stretched to y
a fixed amount that is then held constant while it continues to rotate.
As it rotates, the stress changes from xx-dominated to yy-dominated, so it is clearly F

changing with time, and therefore 0. However, the rate of deformation tensor is
zero because no deformations are taking place, only rotations. So there is a clear
mismatch occurring between the two tensors, which would otherwise be ideally x
suited to each other.
Stress and Strain

Start with the most general form of linearized material behavior. This is
PK2
= C : E
The stiffness tensor, C , is 3x3x3x3 and can represent any material behavior, isotropic or orthotropic, as long as it is linear. The 2nd Piola-
Kirchhoff stress and Green strain tensors are paired together because of their compatibility, i.e., both are defined in the reference
configuration. (It is also convenient that the two are energetic conjugates, although this is not critical if the strains are small.)
The next step is to substitute the transformation from Cauchy stress to 2nd Piola-Kirchhoff stress. This gives
1 T
JF F = C : E
and then solve for the Cauchy stress
1
T
= F (C : E) F
J
Now take the time derivative
1
J T T

= ( ) F (C : E) F + F (C : E) F +
2
J J
1 1 T
T
F (C : E ) F +
F (C : E) F
J J
And substitute in several identities
http://www.continuummechanics.org/corotationalderivative.html 1/5

J T
T T T
tr(D) = F = L F F = F L E = F D F
J
to get
1 1
T T
= tr(D) F (C : E) F + L F (C : E) F +
J J
1 1
T T T T
F (C : (F D F)) F + F (C : E) F L
J J
This big equation can be shortened by recognizing that many of the terms contain J
1
F (C : E) F
T
, which is just the Cauchy stress, .
1
T T T
= tr(D) + L + L + F (C : (F D F)) F
J
And a little more clean-up.
1
T T T
L L = tr(D) + (F F C F F ) : D
J
The term involving tr(D) is usually neglected because the trace is negligibly small in most cases. In fact, it is identically zero for
incompressible materials.
The term involving C represents a rigid body rotation of the stiffness tensor. The constituants include 1
(F F C F
T
F
T
) and are
J
replaced by C .
o
The left hand side (LHS) is called the Lie Derivative and represented by multiply symbols. Two of these are and . Of these, is
o
preferred because is usually used for something else. That something else is the so-called Jaumann derivative, which is very closely
related and will be discussed shortly.
So the entire equation is written simply as

= C : D
If the object is rotating, but not deforming, then D = 0 , and this leaves

T
= L L = 0 when D = 0
Now we see what it takes to compensate for the fact that 0. It is the two terms involving the velocity gradient, L . Furthermore, since
D = 0, then L reduces to W because L = D + W . This leaves

T
= W W = 0 when D = 0
o
But WT = W because W is antisymmetric. And the resulting formula has another specific name. It is the Jaumann derivative, .
o
= W + W = 0 when D = 0
Although the Lie derivative appears to be most general, the Jaumann derivative seems to be the most popular. For example, it is common
to see
o

= C : D
The apostrophe is often dropped, although it is understood from the context that C must in fact be C .
Symbols and Terms

o
It's important not to confuse with , or even . The symbol is still the Cauchy stress, and is still its time derivative. The
terms with the spin tensor simply compensate for the rate of change of the Cauchy stress when rigid body rotations are present.
This topic has many different names. for example, the Jaumann derivative is also called the Jaumann stress rate, or simply the
Jaumann rate. And both the Jaumann derivative and Lie derivative fall under the category of corotational derivatives, or
corotational stress rates, or simply corotational rates.
And finally, there is the issue of objectivity. The rate of deformation tensor is objective because its computation is still correct in
o o
the presence of rotations. is not considered objective because C : D . But is objective because = C : D . So the
terms objective stress rates, or simply objective rates also turn up.
Corotational Rate Example in 2-D
Start with a 2-D rotation matrix, R .
cos sin
R = [ ]
sin cos
so R
is
sin cos

R = [ ]
cos sin
And R T is
T
cos sin
R = [ ]
sin cos
Multiplying the two together gives the spin tensor, W .
T
sin cos cos sin

W = R R = [ ][ ]
cos sin sin cos
0 1
= [ ]
1 0
Now assume a constant stress tensor
20 0
= [ ]
0 0
So as the object rotates (without deforming), the stress tensor evolves as
T
= R R
cos sin 20 0 cos sin

= [ ][ ][ ]
sin cos 0 0 sin cos
2
20 cos 10 sin(2)
= [ ]
2
10 sin(2) 20 sin
And the derivative of the stress tensor is
20 sin(2) 20 cos(2)
= [ ]
20 cos(2) 20 sin(2)
Even though the material is not deforming, the stress tensor is clearly changing with time, and 0. So now compute W
and W in order to calculate the Jaumann derivative.
2
0 1 20 cos 10 sin(2)
W = [ ][ ]
2
1 0 10 sin(2) 20 sin
2
10 sin(2) 20 sin
= [ ]
2
20 cos 10 sin(2)
And
2
20 cos 10 sin(2) 0 1
W = [ ][ ]
2
10 sin(2) 20 sin 1 0
2
10 sin(2) 20 cos
= [ ]
2
20 sin 10 sin(2)
And putting everything together gives
2 2
20 sin(2) 20 cos(2) 10 sin(2) 20 sin 10 sin(2) 20 cos
W + W = [ ] [ ] + [ ]
2 2
20 cos(2) 20 sin(2) 20 cos 10 sin(2) 20 sin 10 sin(2)
0 0
= [ ]
0 0
So it does indeed work.
The key purpose of this example was not to show that one should always get a zero result, because one should not always. The purpose of
this example was to show that if no deformation is taking place at a given instant, then the Jaumann derivative is zero, even though 0.

1 Login
Sort by Oldest
Recommend 2 Share
Join the discussion
LOG IN WITH
Name
John a month ago

Hi,
Good explanation. There is one thing I don't understand though. You wrote that you can replace L's with W's if D = 0, which results in the
commonly used expression for the Jaumann rate of Cauchy stress. However, this expression is generally used in the analyses even if
sigma^triangle /= 0 and D /= 0. What is the point of the replacing L with W if it requires D = 0, i.e. no stretch is allowed?
Reply Share
BobMcG Moderator a month ago

John, I think you understand it exactly right because I think it's confusing too. It appears to be an acceptable approximation to do so and
people choose to live with it. It's probably fine in problems that have large rotations, but small deformations. For example, a metal object
that is rotating, but only deforming elastically (not plastic), would probably have very small D values and large W values. So the difference
between W and L would be acceptably small.
1 Reply Share
Thank you for visiting this webpage. For $4.99, you receive two formatted For $12.95, you receive two formatted
Feel free to email me if you have PDFs (the first for 8.5" x 11" pages, the PDFs (the first for 8.5" x 11" pages, the
questions. second for tablets) of the complete second for tablets) of the entire
stress chapter. website.
Also, please consider visiting an
advertiser above. Doing so helps to Click here to see a sample page in Click here to see a sample page in
each of the two formats. each of the two formats.
cover website hosting fees.
Bob McGinty Email address to receive PDFs Email address to receive PDFs
bmcginty@gmail.com
Table of Contents
Von Mises Stress Equilibrium
12/20/2017 Equilibrium

Equilibrium Search
home > stress > equilibrium
Introduction
This page is all about F = m a, except we will express the forces as stresses acting on differential
sized areas. The first example will be 2-D, to minimize the complexity. Then the equations will be
developed in 3-D, and also presented in cylindrical coordinates.
Following development of the equations, applications will be presented that involve Airy stress functions
and tire mechanics. Finally, the equilibrium equations are used to develop expressions for the speed of
stress waves in steel, aluminum, and rubber.
2-D Equilibrium
The 2-D differential object is shown in the sketch at the right. The idea is to sum all the forces on it and set
them equal to m a. This can be done one component at a time, so start with the x-direction. The forces
consist of
xx acting on face dy in the x

direction
syy + jsyy dy jtxy
acting on face dx in the x
j y txy + dy
xy
direction
j y
txy + jtxy dx
xx +
xx
dx acting on face dy
sxx y
j x
x
in the +x direction
x sxx + jsxx dx
xy
txy j x
xy + dy acting on face dx
y
in the +x direction txy syy

Plus "body forces". These include
any forces due to gravity,
magnetism, etc, and are summarized simply as fx dxdy where fx is force per unit mass
http://www.continuummechanics.org/equilibrium.html 1/17
The mass is density times volume: dxdy .
v x v x
Acceleration is simply ax , although it is perfectly permissible to use the material derivative: t
+ vx
x
.
Summing all this up gives
xx xy
xx dy xy dx + (xx + dx) dy + (xy + dy) dx + fx dxdy = dxdy ax
x y
Cleaning up terms that cancel, and dividing through by dxdy gives
xx xy
+ + fx = ax
x y
And summing forces in the y-direction leads to
yy xy
+ + fy = ay
y x
It is interesting how the equations tie together changes in all the different stress components, making them
interdependent on each other.
An object is said to be in equilibrium when the right hand sides (RHS) of the equations are zero.
3-D Equilibrium
The process in 3-D is the same in principle, only there are more components involved. Performing the
same exercise of summing forces in the x-direction and setting them equal to the x-direction acceleration
goes as follows. (This time, an x1 , x2 , x3 coordinate system is used.)
The forces consist of
11 acting on face dx2 dx3 in the x direction
11
11 + dx 1 acting on face
x 1
dx 2 dx 3 in the +x direction
21
x 2
31
x 3
The body force is fx dx1 dx2 dx3

where fx is force per unit mass
The mass is density times volume:

dx 1 dx 2 dx 3 .
v x v x
Acceleration is simply ax , although it is perfectly permissible to use the material derivative: + vx .
t x
Summing all this up gives
11 21 31
dx 1 dx 2 dx 3 + dx 1 dx 2 dx 3 + dx1 dx2 dx3 + fx = dx1 dx2 dx3 ax
x1 x2 x3
Dividing through by the differential volumes gives
11 21 31
+ + + fx = ax
x1 x2 x3
And since the stress tensor is symmetric...
11 12 13
+ + + fx = ax
x1 x2 x3
The complete set of equations is
11 12 13
+ + + fx = ax
x1 x2 x3
21 22 23
+ + + fy = ay
x1 x2 x3
31 32 33
+ + + fz = az
x1 x2 x3
All this is written in matrix and tensor notation as

t s s tte at a d te so otat o as
+ f = a ij,j + fi = ai
Or one could write the acceleration as the material derivative.
v v
+ f = ( + v ) ij,j + fi = (vi,t + vk vi,k )
t x
Equilibrium in Cylindrical Coordinates

z
The equilibrium equations in cylindrical coordinates contain several additional terms, such as r
and r
,
that further complicate matters.
1 1 r rz
(rrr ) + + + fr = ar
r r r z r
1 1 z r
(rr ) + + + + f = a
r r r z r
1 1 z zz
(rrz ) + + + fz = az
r r r z
Centripetal Acceleration
It is possible to get a quick, rough estimate of the circumferential stress level in a tire undergoing
axisymmetric centripetal forces during a high speed limit test. The radial acceleration equation is
1 1 r rz
(rrr ) + + + fr = ar
r r r z r
The radial acceleration is
2
V
ar =
r
The other terms are expected to be negligible, except /r. Setting these two equal to each
other gives
2
V
=
r r
This simplifies to
2
= V
So for a tire spinning at 200 kph (= 55.55 m/s), with rubber density equal to 1,150 kg/m3, the
circumferential stress should be around
3 2
= (1150kg/m )(55.55m/s) = 3, 500, 000Pa = 3.5MPa
For the steel in the NSTs, the density is 7,800 kg/m3, and the circumferential stress should be
around
3 2
= (7800kg/m )(55.55m/s) = 24, 000, 000Pa = 24MPa
The fact that the tire is actually a nonhomogeneous composite probably makes the actual values
significantly different from these estimates.
Steel Belt Equilibrium
This example relates interply shear strain, xz , which is present between the steel belts and
peaks at the belt edge, to intraply shear stress, xy , in the plane of the belts.
The main governing equilibrium equation for this situation is
xx xy xz
+ + + fx = ax
x y z
Assume that several terms in the equation are negligible, leaving only
xy xz
+ = 0
y z
The interply rubber layer develops shear, called xz . Therefore the shear stress is
xz = Gxz
Now focus on the top belt. The shear stress, xz , in the shear layer is the shear stress on the
bottom surface of the belt. But the shear stress on the top is near zero. So the change in shear
stress through the thickness of the belt is
xz top bottom 0 Gxz G

= = = ( ) xz
z D D D
Substituting this into the equilibrium equation gives
xy G
( ) xz = 0
y D
So the intraply shear in the belt can be related to the interply shear strain as
G
xy = ( ) xz dy
D
Granted, this equation my not be very useful by itself. But it is essential to the general analytical
solution for the stresses and strains in the belts.
Airy Stress Functions

The use of Airy Stress Functions is a powerful technique for solving 2-D equilibrium elasticity problems.
The approach will be presented here for the special case of no body forces.
First, note that in 2-D equilibrium (a = 0 ), and in the absence of body forces (f = 0 ), the equilbrium
equations reduce to
xx xy yy xy
+ = 0 + = 0
x y y x
Next, propose that a scalar function, , exists (the Airy stress function) and is related to the 2-D stress
components by the following cleverly chosen relationships.
2 2 2

xx = yy = xy =
2 2
y x x y
Then, substituting the above relationships into the equilibrium equations gives a remarkable result.
2 2

( ) ( ) = 0
2
x y y x y
2 2

( ) ( ) = 0
2
y x x x y
The remarkable result here is that the equilibrium equations are always satisfied regardless of the choice
of . So any choice of is the solution to a problem (well almost, more on this in a moment). But which
problem? Indeed, when one works with Airy stress functions, one can find oneself with a solution, but not
know what problem it is a solution to!
Take for example, = 12 Ay 2 . This is the solution to something. But what? To find out, take the partial
derivatives to determine the stress fields. This leads to
2
1
2
xx = ( Ay ) = A
2
y 2
Therefore, this is easily recognized as a simple case of uniaxial tension in the x direction. Likewise, letting
= Bxy leads to a state of uniform pure shear in which xy = B.
It is perhaps worth noting that Airy stress functions have been used extensively in the field of Fracture
Mechanics.
Nevertheless, nothing is quite THAT easy. There is one limitation on the choice of that results from the
facts that the solutions are restricted to isotropic materials, the strains are related to stresses through
Hooke's Law, and they must make physical sense, e.g., the strains cannot be so negative that the material
folds back on itself. The limitation is that must satisfy the Biharmonic Equation. It is
4 4 4

+ 2 + = 0
4 2 2 4
x x y y
and is abbreviated 4 = 0. It is not at all intuitive why the restrictions lead to the biharmonic equation,
and there is a great deal of tedious algrebra required to show it, but it is indeed the case. Any function
satisfying 4 = 0 is guaranteed to produce stress and strain fields that are in equilibrium for an isotropic
solid not subjected to body forces.
Note that any polynomial of degree 3 or less in x and y is automatically a solution of the biharmonic
equation because the equation contains 4th order derivatives.
Polar Coordinates
In polar coordinates, the biharmonic equation is
2
2
1 1
( (r ) + ) = 0
2 2
r r r r
and the relationships for the stress components are
2 2
1 1 1
rr = + = r = ( )
2 2 2
r r r r r r
Line Load Example
The case of a distributed linear load

P on an infinite solid can be

)
solved with Airy stress functions in
N/m
(
polar coordinates. The stress P'
function in this case is

P
= r cos
0
r
Th f ti b i t d i th
The function can be inserted in the
biharmonic equation to verify that it
is indeed a solution. The stress
components obtained from
differentiating the stress function
are therefore a valid solution to a
particular problem. But which one? To determine that, first evaluate the stresses.
2
1 1
rr = +
2 2
r r r

2P
= cos
r
2

= = 0
2
r
1
r = ( ) = 0
r r
This stress field results from a distributed line load of zero width. This can be varified by
computing the net vertical force due to the radial stress using
/2
Vertical Load / Length = rr cos rd

/2
where the cos term gives the vertical component of force due to the radial stress. Substituting
the expression for rr into the equation and integrating gives
/2
2P
Vertical Load / Length = ( cos ) cos rd
/2
r
/2
2P
2
= cos d
/2

= P
Equilibrium and the Speed of Stress Waves

It turns out that the equilibrium equation is very useful to the estimation of the speed of stress waves in
materials. The process starts by pulling in a few seemingly unrelated topics. For starters, recall that the
wave equation is
2 2
u 1 u
=
2 2 2
x c t
where u represents displacement, and c is the speed of the stress waves in the material - effectively the
speed of sound in the material. (And this is the focus of this discussion.)
Now bring up the following equilibrium equation
xx xy xz
+ + + fx = ax
x y z
and neglect the shear and body force terms, leaving only
xx
= ax
x
And now substitute several relationships. Begin by noting that ax just like in the wave equation.
u
= 2
t
Next, note that xx in the equilibrium equation is related to xx by
xx = Exx
for the case of uniaxial tension. But then, xx is related to the displacements through
u
xx =
x
Again, this is for the simple case of uniaxial tension. So stress can be related to displacements by
u
xx = Exx = E
x
And
2
xx u
= E
2
x x
Substituting all this into the equilibrium equation gives
2 2
u u
E =
2 2
x t
Or
2 2
u 1 u
=
2 2
x (E/) t
Now the big finish.... Comparing this to the wave equation shows that

E E
2
c = or c =

And that is the relationship for the speed of a uniaxial stress wave through a material, its speed of sound!
Speed of Sound in Materials

3
For steel, E = 200(10) Pa
9
and = 7, 800kg/m . So this gives

9
E 200(10) Pa
c = = = 5km/s = 5m/ms
3

7, 800kg/m
3
For aluminum, E = 70(10) Pa
9
and = 2, 800kg/m . So this gives

9
E 70(10) Pa
c = = = 5km/s = 5m/ms
3
2, 800kg/m
By coincidence, the speed of sound through both steel and aluminum is the same.
3
For rubber with E 6
= 1(10) Pa and = 1, 150kg/m . So this gives

6
E 1(10) Pa
c = = = 29m/s = 0.03m/ms
3

1, 150kg/m
Shear Wave Speeds

But we're not done! Take a look at shear waves. This time, bring up the following equilibrium equation
yx yy yz
+ + + fy = ay
x y z
and neglect all the terms except
yx
= ay
x
And swap the subscripts on since it's symmetric.

2
Begin by substituting ay =
v
2
just like in the wave equation.
t
And relate xy to xy by
xy = Gxy
And relate xy to the displacements with the simple-shear assumption.
v
xy =
x
So the shear stress can be related to displacements by
v
xy = Gxy = G
x
And
2
xy v
= G
2
x x
Substituting all this into the equilibrium equation gives
2 2
v v
G =
2 2
x t
Or
2 2
v 1 v
=
2 2
x (G/) t
Comparing this to the wave equation shows that

G G
2
c = or c =

So for shear waves their speed depends on the shear modulus, G, not the elongation modulus, E. For
incompressible materials, i.e., rubber, the shear modulus is one-third of the tension modulus, so shear

waves propagate through rubber at 1/3 , or 58% of the speed of uniaxial tension waves. For metals, the
shear modulus is about 38% of the tension modulus. This translates to their shear wave speeds being 61%
of their tension wave speeds.
Plane Wave Speeds

And finally, there are plane wave speeds. These are cases where the cross-sections of the objects are
very large and hold the lateral strains constant at zero while the object undergoes tension/compression.
This doesn't really apply to rubber because volume changes are involved.
To see what happens, go back to Hooke's Law for stress in terms of strain
E
ij = [ij + ij kk ]
(1 + ) (1 2 )
and impose the following: 11 = , and 22 = 33 = 0 . This gives
E
= [ + ]
(1 + ) (1 2 )
which simplifies to
E (1 )
=
(1 + )(1 2 )
You can see that for rubber with = 0.5, the stress required to generate any strain is infinite due to the
(1 2 ) term in the denominator. This is because rubber is incompressible.
As before, substitute for . This gives

u
E (1 ) u
=
(1 + )(1 2 ) x
And
2
E (1 ) u
=
2
x (1 + )(1 2 ) x
Combining everything gives
2 2
E (1 ) u u
=
2 2
(1 + )(1 2 ) x t
So the wave speed of plane waves is

(1 ) E
c = ( )
(1 + )(1 2 )
So for metals with = 1/3, the plane wave speed is 22% greater than the uniaxial tension case. And for
incompressible rubber with = 1/2, the speed would theoretically be infinite, which is of course
impossible. This occurs because compression from a plane wave must result in a volume change, which
would theoretically propagate at infinite speed in an incompressible material.
This is resolved by recalling from Hooke's Law that

rubber is actually compressible and its bulk modulus is approximately 1,000 MPa. For the plane wave,
xx = K xx because yy = zz = 0. This leads to

K
c =

This produces a plane wave speed in rubber of approximately

K 1, 000E6Pa
c = = = 930m/s = 0.93m/ms
3

1, 150kg/m
Keep in mind that this is only an approximation based on an estimate of the bulk modulus. Nevertheless,
the speed is still much less than that in metals.

1 Login
Sort by Oldest
Recommend Share
Join the discussion
LOG IN WITH
Name
Sanjay Pandey 3 months ago

my first question is, in the equilibrium equation, xx/x, x in x is X(reference) or x (deformed).
second question: in topic Equilibrium and the Speed of Stress Waves, the x in xx=u/x is X(reference)
or x (deformed). because in in deformation gradient topic x in x is X(reference). please clarify it. it will be
better to follow consistence notations.
And my third question : what is u here? Is it x-X ?
Reply Share

The equations are in x (deformed) coordinates because that is the state the object would be in for
ilib i It i ti ilib i
http://www.continuummechanics.org/equilibrium.html
i th f t t b if it it ld 't dt 15/17
equilibrium. It is not in equilibrium in the reference state because if it were, it wouldn't need to
move/deform, but it obviously does, even if only a tiny amount. You are right about u = x - X.
Reply Share
Sanjay Pandey > BobMcG 3 months ago

thanx for your prompt reply, but still i have some question that in the equation xx=Exx=Eu/x,
if x is reference not deformed as per strain formulation by deformation gradient (x/X = X/X +
u/X). here strain is derived from u/X(reference).
if my above statement is true than how E2u/x2 = 2u/t2 is true because here x deformed as
per your statements. please reply. thanx
Reply Share

The equilibrium equations are good in any general case. However, Hooke's Law (xx=Exx) is only
useful when strains are relatively small. Usually so small that differences between the reference and
deformed configurations are negligible, so it doesn't matter which one is used: "X" or "x". Also, metals will
yield when strains are large and no longer follow xx=Exx anyway. So the wave equations derived here
really are only good for small strains.
Reply Share
hosting fees.
formats. formats.
Bob McGinty
bmcginty@gmail.com
Table of Contents
Corotational Derivatives Material Behavior

MMK Bab 5

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

MMK Bab 5

Uploaded by

Copyright:

Available Formats

12/20/2017 Stress Introduction

Search Continuum Mechanics Website

home > stress > stress intro

When the force vector is parallel to the surface, the

Many components are needed to capture a

txy Rotational txy

txy x sxx x sxx

But since xy = yx , all the tensors can also be written as

0 Comments Continuum Mechanics

Start the discussion

Be the first to comment.

Subscribe d Add Disqus to your siteAdd DisqusAdd Privacy

Thank You Stress Chapter - $4.99 Entire Website - $12.95

Copyright 2012 by Bob McGinty

Search Continuum Mechanics Website

home > stress > traction vector

So T has units of stress, like MPa, but it is

Calculating a Traction Vector

It is cut (virtually) so the traction vector is

Note the units are MPa now rather than N.

This time, the object is cut (virtually again) at a 30 angle.

The applicable area in this case is

And the traction vector is

Normal and Shear Stresses

Normal and Shear Stress from a Traction Vector

The unit normal to the surface is

So the normal stress on the surface is

The vector parallel to the surface is

The shear stress on the surface is

Stress Tensors and Traction Vectors

xx A cos + xy A sin = Tx A sxx Ty

Replacing the cos and sin with n x and n y gives

Traction Vector from Stress Tensor

Calculate the traction vector on a surface with unit normal n .

So T = 46.79 i + 74.86 j + 34.40 k MPa .

Substituting this equation for T into the above ones gives

In tensor notation, the equations are

Stress Transform Example

Recall the above stress tensor

We had calculated the traction vector on a surface with unit normal

The normal stress on the surface is

{0.400 0.600 0.693} 50 10 30 0.400

{0.832 0.555 0.000} 50 10 30 0.400

n s = (0.400 i + 0.600 j + 0.693 k) (0.832 i + 0.555 j + 0.000 k)

= 0.385 i 0.576 j + 0.721 k

So the shear in the direction perpendicular to the first is

{-0.385 -0.576 0.721} 50 10 30 0.400

case for strain tensors.

Forces on Cross Sections

Both forms turn up often in literature.

2 Comments Continuum Mechanics

Join the discussion

BobMcG Moderator > Brajesh Kumar 3 months ago

Subscribe d Add Disqus to your siteAdd DisqusAdd Privacy

Thank You Stress Chapter - $4.99 Entire Website - $12.95

Copyright 2012 by Bob McGinty

Search Continuum Mechanics Website

home > stress > energetic conjugates

Cauchy Stress (a.k.a. True Stress)

Cauchy stress is "force over area in the

Cauchy stress is represented simply by . All

Work, Energy, and Power

But dx/dt is simply velocity, v . So power is