Vous êtes sur la page 1sur 1301

Multimedia course

CONTINUUM MECHANICS
FOR ENGINEERS

By Prof. Xavier Oliver


Technical University of Catalonia (UPC/BarcelonaTech)
International Center for Numerical Methods in Engineering (CIMNE)
http://oliver.rmee.upc.edu/xo
First edition May 2017 This material is distributed under the terms of the Creative Commons Attribution Non-
Commercial No-Derivatives (CC-BY-NC-ND) License, which permits any noncommercial use,
DOI: distribution, and reproduction in any medium of the unmodified original material ,provided the
original author(s) and source are credited.
The available media
 786 slides  550 book-pages

 141 movies
How to interact with the media ...
1 Choose the slide with the subject you are interested in
How to interact with the media ...
2 To watch the movie explaining the slide, click on
the link-to-YouTube icon
How to interact with the media ...
3 To reach the book-section dealing with the slide,
click on the link-to-book icon
How to interact with the media ...
4 To return to the slide, click on the link-to-slide icon
Acknowledgements
 Prof. Carlos Agelet (CIMNE/UPC)
 Dr. Manuel Caicedo (CIMNE/UPC)
 Dr. Eduardo Car (CIMNE)
 Prof. Eduardo Chaves (UCLM)
 Dr. Ester Comellas (CIMNE)
 Dr. Alex Ferrer (CIMNE/UPC)
 Prof. Alfredo Huespe (CIMNE/UNL/UPC)
 Dr. Oriol Lloberas-Valls (CIMNE/UPC)
 Dr. Julio Marti (CIMNE)

… and the past students of my courses on


Continuum Mechanics …

Thanks for your contribution !!!!


CONTINUUM MECHANICS FOR ENGINEERS
Course contents
 General Principles
 Ch.1. Description of Motion
 Ch.2. Deformation and Strain
 Ch.3. Compatibility Equations
 Ch.4. Stress
 Ch.5. Conservation and Balance Equations
 Constitutive Equations
 Ch.6. Linear Elasticity
 Ch.7. Plane Linear Elasticity
 Ch.8. Plasticity
 Ch.9. Constitutive Equations in Fluids
 Ch.10. Fluid Mechanics
 Additional Content
 Ch.11. Variational Principles
 Appendix
 Tensor Algebra
CH.1. DESCRIPTION OF
MOTION
Multimedia Course on Continuum Mechanics
Overview
1.1. Definition of the Continuous Medium
1.1.1. Concept of Continuum Lecture 1
1.1.2. Continuous Medium or Continuum

1.2. Equations of Motion


1.2.1 Configurations of the Continuous Medium
1.2.2. Material and Spatial Coordinates Lecture 2
1.2.3. Equation of Motion and Inverse Equation of Motion
1.2.4. Mathematical Restrictions
1.2.5. Example Lecture 3
1.3. Descriptions of Motion
1.3.1. Material or Lagrangian Description Lecture 4
1.3.2. Spatial or Eulerian Description
1.3.3. Example Lecture 5

2
Overview (cont’d)
1.4. Time Derivatives Lecture 6
1.4.1. Material and Local Derivatives
1.4.2. Convective Rate of Change Lecture 7
1.4.3. Example Lecture 8

1.5. Velocity and Acceleration


1.5.1. Velocity
Lecture 9
1.5.2. Acceleration
1.5.3. Example

1.6. Stationarity and Uniformity


1.6.1. Stationary Properties Lecture 10
1.6.2. Uniform Properties

3
Overview (cont’d)
1.7. Trajectory or Pathline
1.7.1. Equation of the Trajectories
1.7.2. Example
1.8. Streamlines Lecture 11
1.8.1. Equation of the Streamlines
1.8.2. Trajectories and Streamlines
1.8.3. Example
1.8.4. Streamtubes

1.9. Control and Material Surfaces


1.9.1. Control Surface
1.9.2. Material Surface Lecture 12
1.9.3. Control Volume
1.9.4. Material Volume

4
1.1 Definition of the Continuous
Medium
Ch.1. Description of Motion

5
The Concept of Continuum
 Microscopic scale:
 Matter is made of atoms which may be grouped in
molecules.
 Matter has gaps and spaces.

 Macroscopic scale:
 Atomic and molecular discontinuities are disregarded.
 Matter is assumed to be continuous.

6
Continuous Medium or Continuum
 Matter is studied at a macroscopic scale: it completely
fills the space, there exist no gaps or empty spaces.

 Assumption that the medium and is made of infinite


particles (of infinitesimal size) whose properties are
describable by continuous functions with continuous
derivatives.

7
Exceptions to the Continuous Medium

 Exceptions will exist where the theory will not account


for all the observed properties of matter. E.g.: fatigue
cracks.
 Inoccasions, continuum theory can be used in combination
with empirical information or information derived from a
physical theory based on the molecular nature of material.

 Theexistence of areas in which the theory is not


applicable does not destroy its usefulness in other
areas.

8
Continuum Mechanics
 Study of the mechanical behavior of a continuous
medium when subjected to forces or displacements, and
the subsequent effects of this medium on its
environment.

 It divides into:
 General Principles: assumptions and consequences
applicable to all continuous media.
 Constitutive Equations: define the mechanical behavior of
a particular idealized material.

9
1.2 Equations of Motion
Ch.1. Description of Motion

10
Material and Spatial points,
Configuration
 A continuous medium is formed by an infinite number of
particles which occupy different positions in space during
their movement over time.
 MATERIAL POINTS: particles
 SPATIAL POINTS: fixed spots in space

 The CONFIGURATION Ωt of a continuous medium at a


given time (t) is the locus of the positions occupied by the
material points of the continuous medium at the given time.

11
Configurations of the Continuous
Medium
Ω0: non-deformed (or reference) Ω or Ωt: deformed (or present)
configuration, at reference time t0. configuration, at present time t.
Γ0 : non-deformed boundary. Γ or Γt : deformed boundary.
X : Position vector of a particle at
reference time. x : Position vector of the same
particle at present time.
ϕ ( X,t ) Γ
t0= 0 → reference time t ∈ [ 0, T ] → current time
Γ0

Initial, reference Ω
or undeformed
configuration
Ω0 X x
Present or deformed
configuration

12
Material and Spatial Coordinates
 The
position vector of a given particle can be
expressed in:
 Non-deformed or Reference Configuration
 X1  X 
[ X]  X=
= 2
  
 Y  ≡ material coordinates (capital letter)
X  Z 
 3  

 Deformed or Present Configuration


 x1  x
[ x]  x=
= 2
  
 y  ≡ spatial coordinates (small letter)
x  z
 3  

13
Equations of Motion
 The motion of a given particle is described by the evolution
of its spatial coordinates (or its position vector) over time.
= x ϕ= (particle label, t ) x ( particle label, t )

=  xi ϕi ( particle label, t ) i ∈ {1, 2,3}

φ (particle label, t) is the motion that takes


the body from a reference configuration to
the current one.

 The Canonical Form of the Equations of Motion is obtained


when the “particle label” is taken as its material coordinates
 not
= x ϕ= ( X, t ) x ( X, t )
particle label ≡ { X 1 X 2 X 3 } ≡ X
T

=  xi ϕi ( X 1 , X 2 , X 3 , t ) i ∈ {1, 2,3}

14
Inverse Equations of Motion
 The inverse equations of motion give the material
coordinates as a function of the spatial ones.
ϕ ( X,t ) Γ
Γ0

Ω0 X ϕ −1 ( x, t ) x

 not
=  X ϕ= −1
( x, t ) X ( x, t )

= X i ϕi −1 ( x1 , x2 , x3 , t ) i ∈ {1, 2,3}

15
Mathematical restrictions for φ and φ-1
defining a “physical” motion
 Consistency condition
 ϕ ( X, 0 ) = X , as X is the position vector for t=0
 Continuity condition
ϕ ∈ C , φ is continuous with continuous derivatives
1

 Biunivocity condition
 φ is biunivocal to guarantee that two particles do not occupy
simultaneously the same spot in space and that a particle does not
occupy simultaneously more than one spot in space.
Mathematically: the “Jacobian” of the motion’s equations should
be different from zero: ∂ϕ ( X, t )  ∂ϕi 
= J = det  ≠0 
∂X  ∂X j 

 The “Jacobian” of the equations of motion should be


“strictly positive” ∂ϕ ( X, t )  ∂ϕ  density is always positive
= J = det  >0
i

∂X  ∂X j  (to be proven)
16
Example
The spatial description of the motion of a continuous medium is given by:

=  x1 X= e 2t
 x Xe 2t


1

x ( X, t=
) ≡  x2 X 2e−2t = ≡  y Ye −2t
x = + 2t z = + 2t
 3 5 X 1t X 3 e  5 Xt Ze

Find the inverse equations of motion.

17
 x1 = X 1e 2t

Example - Solution 
x ( X, t ) ≡  x2 = X 2 e −2t
 x 5 X t + X e 2t
=
 3 1 3

Check the mathematical restrictions:


 Consistency Condition ϕ ( X, 0 ) = X ?
 X 1e 2⋅0   X1 
   
x ( X,= )  X 2e−2⋅0 =
t 0=  X 2= X
5 X ⋅ 0 + X e  2 ⋅ 0 X 
 1 3   3
 Continuity Condition ϕ ∈ C1 ?
 Biunivocity Condition ?
∂x1 ∂x1 ∂x1
∂X 1 ∂X 2 ∂X 3
e 2t 0 0
∂xi ∂x ∂x2 ∂x2
J= = 2 = 0 e −2t 0 = e 2t ⋅ e −2t ⋅ e 2t = e 2t ≠ 0 ∀t
∂X j ∂X 1 ∂X 2 ∂X 3
5t 0 e 2t
∂x3 ∂x3 ∂x3
∂X 1 ∂X 2 ∂X 3
∂ϕ ( X, t )
 Density positive ? =J >0
=
J e 2t > 0
∂X

18
 x1 = X 1e 2t

Example - Solution 
x ( X, t ) ≡  x2 = X 2 e −2t
 x 5 X t + X e 2t
=
 3 1 3

Calculate the inverse equations:


x
x1 =X 1e 2t ⇒ X 1 = 21t =x1e −2t
e
x
x2 =X 2 e −2t ⇒ X 2 = −22t =x2 e 2t
e
x − 5X t
x3 =
5 X 1t + X 3e 2t ⇒ X 3 =3 2t 1 =
e
( )
x3 − 5 ( x1e −2t ) t e −2t =
x3e −2t − 5tx1e −4t

 X 1 = x1e −2t

X ≡ ϕ −1 ( x, t ) =
X2 = x2 e 2t
X
= −2 t −4 t
 3 x3e − 5tx1e

19
1.3 Descriptions of Motion
Ch.1. Description of Motion

20
Descriptions of Motion
 Themathematical description of the particle properties
can be done in two ways:

 Material (Lagrangian) Description

 Spatial (Eulerian) Description

21
Material or Lagrangian Description
 The physical properties are described in terms of the
material coordinates and time.

 It focuses on what is occurring at a fixed material point (a


particle, labeled by its material coordinates) as time
progresses.

 Normally used in solid mechanics.

22
Spatial or Eulerian Description
 The physical properties are described in terms of the spatial
coordinates and time.

 It focuses on what is occurring at a fixed point in space (a


spatial point labeled by its spatial
coordinates) as time progresses.

 Normally used in fluid mechanics.

23
Example
The equation of motion of a continuous medium is:
=x X − Yt

x =x ( X, t ) ≡  y =Xt + Y
z = − Xt + Z

Find the spatial description of the property whose material description is:
X +Y + Z
ρ ( X,Y,Z,t ) =
1+ t2

24
=x X − Yt

Example - Solution x =x ( X, t )

≡  y =Xt + Y
z =
 − Xt + Z

Check the mathematical restrictions:


 Consistency Condition φ (X,0 ) = X ?
X −Y ⋅0 X 
   
x ( X, t = 0 ) =  X ⋅ 0 + Y  = Y  = X
X ⋅0 + Z  Z 
   
 Continuity Condition φ ∈ C 1 ?
 Biunivocity Condition ?
∂x ∂x ∂x
∂X ∂Y ∂Z 1 −t 0
∂xi ∂y ∂y ∂y
J= = = t 1 0 = 1 ⋅1 ⋅1 − 1 ⋅ ( −t ) ⋅ t = 1 + t 2 ≠ 0 ∀t
∂X j ∂X ∂Y ∂Z
−t 0 1
∂z ∂z ∂z
∂X ∂Y ∂Z
∂φ (X, t )
 Any diff. Vol. must be positive J=
∂X
>0 ?

J =1 + t 2 > 0

25
=x X − Yt

Example - Solution x =x ( X, t )

≡  y =Xt + Y
z =
 − Xt + Z

Calculate the inverse equations:


x=X − Yt ⇒ X=x + Yt 
 y −Y y − xt
y −Y ⇒ x + Yt = ⇒ Yt 2 + Y = y − xt ⇒ Y=
y =Xt + Y ⇒ X=  t 1+ t2
t

 y − xt  x + xt 2 + yt − xt 2 x + yt
X =x + Yt =x +  2 
t= =
 1 + t  1 + t 2
1+ t2

 x + yt  z + zt 2 + xt + yt 2
z =− Xt + Z ⇒ Z =z + Xt =z +  2 
t=
 1+ t  1+ t2
 x + yt
 X =
1+ t2

 y − xt
X ≡ ϕ ( x, t ) =
−1
 Y =2
 1+ t
 z + zt 2 + xt + yt 2
Z =
 1+ t2

26
Example - Solution
Calculate the property in its spatial description:
 x + yt
 X =
1+ t2

 y − xt
X ≡ ϕ ( x, t ) =
−1
 Y =2
 1+ t
 z + zt 2 + xt + yt 2
Z =
 1+ t2

 x + yt   y − xt   z + zt + xt + yt 
2 2

 +
2  
+
2   
X + Y + Z  1+ t   1+ t   1+ t2  x + y + yt + yt 2 + z + zt 2
( X,Y,Z,t ) =
ρ= =
1+ t2 1+ t2 (1 + t 2 )
2

X +Y + Z x + y (1 + t + t 2 ) + z (1 + t 2 )
ρ ( X,Y,Z,t ) = ⇒ ρ ( x, y,z,t )
=
1+ t2 (1 + t )2 2

27
1.4 Time Derivatives
Ch.1. Description of Motion

28
Material and Local Derivatives
 The
time derivative of a given property can be defined
based on the:
 Material Description Γ(X,t) TOTAL or MATERIAL DERIVATIVE
 Variation of the property w.r.t. time following a specific particle
in the continuous medium.
∂Γ ( X,t ) partial time derivative of the
 material derivative ≡ →
∂t material description of the propery

 Spatial Description γ(x,t)  LOCAL or SPATIAL DERIVATIVE


 Variation of the property w.r.t. time in a fixed spot of space.
∂γ ( x,t ) partial time derivative of the
 local derivative ≡ →
∂t spatial description of the propery

29
Convective Derivative
 Remember: x=x(X,t), therefore, γ(x,t)=γ(x(X,t),t)=Γ(X,t)
 The material derivative can be computed in terms of
spatial descriptions:
d
not not D ∂Γ( X, t )
material derivative = → γ ( x, t )
= =γ ( x, t ) =
dt Dt ∂t
∂γ ( x, t ) ∂γ ∂xi ∂γ ( x, t ) ∂γ ∂x
γ ( x ( X, t ) ,=
t)
d
= + ⋅ = + ⋅ =
∂t ∂x ∂t ∂t ∂x ∂t
i   
dt

∇ γ   ∂x  ∇γ ( x,t ) v ( x,t )
   ∂t   
v (x,t )⋅∇γ (x,t )
i  i

 Generalising for any property:


REMARK
d χ ( x, t ) ∂χ ( x, t ) The spatial Nabla operator
= + v ( x, t ) ⋅∇χ ( x, t ) is defined as: ∇ ≡ ∂ ê
dt ∂t
∂xi
i

material local convective rate of


derivative derivative change
30 (convective derivative)
Convective Derivative
 Convective rate of change or convective derivative is
implicitly defined as:
v ⋅∇ ( • )

 The term convection is generally applied to motion related


phenomena.
 If there is no convection (v=0) there is no convective rate of change
and the material and local derivatives coincide.
d (•) ∂ (•)
v ⋅∇ ( • ) 0=
=
dt ∂t

31
Example
Given the following equation of motion:
 x = X + Yt + Zt

x(X, t ) ≡  y = Y + 2 Zt
 z = Z + 3 Xt

And the spatial description of a property ρ(x, t ) = 3x + 2 y + 3t ,


Calculate its material derivative.

Option #1: Computing the material derivative from material descriptions


Option #2: Computing the material derivative from spatial descriptions

32
 x = X + Yt + Zt

Example - Solution
x(X, t ) ≡  y = Y + 2 Zt
 z = Z + 3 Xt

ρ(x, t ) = 3x + 2 y + 3t

Option #1: Computing the material derivative from material


descriptions
Obtain ρ as a function of X by replacing the Eqns. of motion into ρ(x,t) :
ρ ( x, t )= ρ ( x( X, t ), t )= ρ ( X, t )= 3 ( X + Yt + Zt ) + 2 (Y + 2Zt ) + 3t
= 3 X + 3Yt + 2Y + 7 Zt + 3t
Calculate its material derivative as the partial derivative of the material
description:
d ρ ( x, t ) ∂ρ ( X, t ) ∂
= = ( 3 X + 3Yt + 2Y + 7 Zt + 3t ) = 3Y + 7 Z + 3
dt ∂ t ∂t
( )
x = x X,t

d ρ ( x, t ) ∂ρ ( X, t )
= =3 + 3Y + 7 Z
dt ∂t
( )
x = x X,t

33
 x = X + Yt + Zt

Example - Solution
x(X, t ) ≡  y = Y + 2 Zt
 z = Z + 3 Xt

ρ(x, t ) = 3x + 2 y + 3t

Option #2: Computing the material derivative from spatial


descriptions
d ρ ( x, t ) ∂ρ ( x, t )
= + v ( x, t ) ⋅ ∇ρ ( x, t )
dt ∂t
Applying this on ρ(x,t) :
 ∂ρ ( x, t ) ∂
 = ( 3x + 2 y + 3t=) 3
 ∂t ∂ t
 Y + Z 
∂x  ∂ ∂ ∂
T
   
 v ( x, t ) x=x( X,t ) = ∂t =  ∂t ( X + Yt + Zt ) , ∂t (Y + 2 Zt ) , ∂t ( Z + 3 Xt )  = [Y + Z , 2 Z , 3 X ] =  2 Z 
T

  3 X 


( ) ( ) ( )
T T
∇ρ ( x=  ∂ρ x , t ∂ρ x , t ∂ρ x , t   ∂ ∂ ∂ 
,t)  , , =  ∂x ( 3 x + 2 y + 3t ) , ∂y ( 3 x + 2 y + 3t ) , ∂z ( 3 x + 2 y + 3t )=

  ∂ x ∂ y ∂ z   

 3

= [3,= 2, 0]  2 
T

  0 
34
 x = X + Yt + Zt

Example - Solution
x(X, t ) ≡  y = Y + 2 Zt
 z = Z + 3 Xt

ρ(x, t ) = 3x + 2 y + 3t

Option #2
The material derivative is obtained:

3 
d ρ ( x, t )
=3 + [Y + Z , 2 Z , 3 X ]  2  =3 + 3Y + 3Z + 4 Z
dt 
x = x( X,t )  v
T
 0 
  

ρ 

v⋅(ρ )

d ρ ( x, t )
=3 + 3Y + 7 Z
dt x=x X,t
( )

35
1.5 Velocity and Acceleration
Ch.1. Description of Motion

36
Velocity
 Time derivative of the equations of motion.
 Material description of the velocity: REMARK
Time derivative of the equations of motion Remember the
 ∂x ( X, t ) equations of motion
 (
V X, t ) = are of the form:
∂t

not

V X, t ∂xi ( X, t ) = ( X, t ) x ( X, t )
x ϕ=
 i ( )
= i ∈ 1, 2,3
∂t

 Spatial description of the velocity:


Velocity is expressed in terms of x using the inverse equations of
motion:
V ( X ( x, t ) , t ) v ( x, t )

37
Acceleration
 Material time derivative of the velocity field.
 Material description of acceleration:
Derivative of the material description of velocity:
 ∂V ( X, t )
 (
A X, t ) =
∂t

 A X, t ∂Vi ( X, t )
 i ( )
= i ∈ 1, 2,3
∂t
 Spatial description of acceleration:
A(X,t) is expressed in terms of x using the inverse equations of motion:
A ( X ( x, t ) , t ) a ( x, t )
Or a(x,t) is obtained directly through the material derivative of v(x,t):
 dv ( x, t ) ∂v ( x, t )
 =
a ( x, t ) = + v ( x, t ) ⋅∇v ( x, t )
 dt ∂t

a ( x, t ) = dvi ( x, t ) = ∂vi ( x, t ) + v ( x, t ) ⋅ ∂vi ( x, t ) i ∈ 1, 2,3
 i dt ∂t
k
∂xk
38
Example
Consider a solid that rotates at a constant angular velocity ω and has the
following equation of motion:

=  x R sin (ωt + φ)

x ( R, φ , t ) → 
label of =  y R cos (ωt + φ)

particle
→ (non - canonical equations of motion)

Find the velocity and acceleration of the movement described in both, material
and spatial forms.

39
 x = R sin(ωt + φ)
Example - Solution 
 y = R cos(ωt + φ )

Using the expressions sin ( a ± b )= sin a ⋅ cos b ± cos a ⋅ sin b



cos ( a ± b )= cos a ⋅ cos b  sin a ⋅ sin b

The equation of motion can be rewritten as:


 x R sin ( ω=
= t + φ ) R sin ( ωt ) cos φ + R cos ( ωt ) sin φ

= y R cos ( ω=t + φ ) R cos ( ωt ) cos φ − R sin ( ωt ) sin φ

For t=0, the equation of motion becomes:=


 X R sin φ

=
Y R cos φ
Therefore, the equation of motion in terms of the material coordinates is:
=Y =X
=x R sin ( ω=
t + φ ) R sin ( ωt ) cos φ + R cos ( ωt ) =
sin φ X cos ( ωt ) + Y sin ( ωt )
=y R cos ( ω=
t + φ ) R cos ( ωt ) cos φ − R sin ( ωt ) =
sin φ − X sin ( ωt ) + Y cos ( ωt )
=Y =X

40
=x X cos ( ωt ) + Y sin ( ωt )
Example - Solution y =− X sin ( ωt ) + Y cos ( ωt )

The inverse equation of motion is easily obtained


x − Y sin ( ωt )
=x X cos ( ωt ) + Y sin ( ωt ) X=
cos ( ωt ) x − Y sin ( ωt ) − y + Y cos ( ωt )
=
− y + Y cos ( ωt ) cos ( ωt ) sin ( ωt )
y =− X sin ( ωt ) + Y cos ( ωt ) X=
sin ( ωt )

x sin ( ωt ) − Y sin 2 ( ωt ) =− y cos ( ωt ) + Y cos 2 ( ωt )

( ωt ) Y ( cos 2 ( ωt ) + sin 2 ( ωt ) )
x sin ( ωt ) + y cos=

=1

x − ( x sin ( ωt ) + y cos ( ωt ) ) sin ( ωt ) x x sin 2 ( ωt ) y cos ( ωt ) sin ( ωt )


X= = − − =
cos ( ωt ) cos ( ωt ) cos ( ωt ) cos ( ωt )
1 − sin 2 ( ωt )
= x − y sin ( ωt )
cos ( ωt )

= cos( ωt )
41
Example - Solution
So, the equation of motion and its inverse in terms of the material coordinates
are:

= x X cos ( ωt ) + Y sin ( ωt )


x ( X, t ) →  → canonical equations of motion
 y =− X sin ( ωt ) + Y cos ( ωt )

=X x cos ( ωt ) − y sin ( ωt )


X ( x, t ) →  → inverse equations of motion
Y x sin ( ωt ) + y cos ( ωt )
=

42
= x X cos ( ωt ) + Y sin ( ωt )
Example - Solution x ( X, t ) → 
 y =− X sin ( ωt ) + Y cos ( ωt )

∂x ( X, t )
Velocity in material description is obtained from V ( X, t ) =
∂t

 ∂x ∂
=
∂x ( X, t )  ∂t ∂t
( X cos ( ωt ) + Y sin ( ωt ) )
V ( X, t ) = 
=
∂t  ∂y = ∂
( − X sin ( ωt ) + Y cos ( ωt ) )
 ∂t ∂t

Vx  − X ω sin ( ωt ) + Y ω cos ( ωt ) 


V (X
=, t ) =   
Vy   − X ω cos ( ω t ) − Y ω sin ( ω t ) 

43
=X x cos ( ωt ) − y sin ( ωt )

Y x sin ( ωt ) + y cos ( ωt )
=
Example - Solution  x X cos ( ωt ) + Y sin ( ωt )
=

 y =− X sin ( ωt ) + Y cos ( ωt )

Velocity in spatial description is obtained introducing X(x,t) into V(X,t):



y
 x X cos ( ωt ) + Y sin ( ωt )
=  v x   − X sin ( ωt ) + Y cos ( ωt )  ω  ωy 
 → v (
= x , t ) V ( X ( x , =
t ) , t ) =  =  
 y =− X sin ( ω t ) + Y cos ( ω t )  v y  
 −

X cos ( ωt ) − Y

sin ( ω t )  ω  − ωx 
−x
Alternative procedure (longer):
−ω x cos ( ωt ) sin ( ωt ) + ωy sin ( ωt ) + ωx sin ( ωt ) cos ( ωt ) + ωy cos ( ωt ) 
 2 2

v ( x, t ) = 

 −ω x cos 2
( ω t ) + ω y sin ( ω t ) cos ( ω t ) − ω x sin 2
( ω t ) − ω y cos ( ω t ) sin ( ω t ) 

ω x ( sin ( ωt ) cos ( ωt ) − cos ( ωt ) sin ( ωt ) ) + ωy ( sin 2 ( ωt ) + cos 2 ( ωt ) ) 
   
 =0 =1 
= 
 −ω x () ()
sin 2
( ωt ) + cos 2
( ω t ) + ω y sin ( ω t ) cos ( ω t ) − cos ( ω t ) sin ( ω t )
 
 =1 =0 

 v x   ω y 
v (=
x, t ) =   
 v y  −ω x 
44
− X ω sin ( ωt ) + Y ω cos ( ωt ) 
Example - Solution V ( X, t ) = 
 − X ω cos ( ω t ) − Y ω sin ( ω t )



∂V ( X, t )
Acceleration in material description is obtained applying: A ( X, t ) =
∂t

 ∂Vx
∂V ( X, t )  ∂t
=− X ω2
cos ( ω t ) − Y ω2
sin ( ωt )
A ( X, t ) = 
=
∂t  ∂Vy = X ω2 sin ( ωt ) − Y ω2 cos ( ωt )
 ∂t

 Ax   X cos ( ωt ) + Y sin ( ωt ) 
2 
A ( X, t ) =   = −ω  
 Ay   − X sin ( ω t ) + Y cos ( ω t ) 

45
X x cos ( ωt ) − y sin ( ωt )
Example - Solution =

Y x sin ( ωt ) + y cos ( ωt )
=

(OPTION #1) Acceleration in spatial description is obtained by replacing the


inverse equation of motion into A(X,t):

( X ( x, t ) , t )
a ( x, t ) A=
=
( x cos ( ωt ) − y sin ( ωt ) ) cos ( ωt ) + ( x sin ( ωt ) + y cos ( ωt ) ) sin ( ωt ) 
= −ω  2

− ( x cos ( ωt ) − y sin ( ωt ) ) sin ( ωt ) + ( x sin ( ωt ) + y cos ( ωt ) ) cos ( ωt ) 

 x ( cos 2 ( ωt ) + sin 2 ( ωt ) ) + y ( − sin ( ωt ) cos ( ωt ) + cos ( ωt ) sin ( ωt ) ) 


    
 =1 =0 
a ( x, t ) = −ω2  
 x ( − cos ( ωt ) sin ( ωt ) + sin ( ωt ) cos ( ωt ) ) + y ( sin ( ωt ) + cos ( ωt ) ) 
2 2

    
 =0 =1 

ax  −ω x 
2

a (=
x, t ) =   2 
a y  −ω y 

46
 v x   ω y 
Example - Solution v (=
x, t ) =   
 v y  −ω x 

(OPTION #2): Acceleration in spatial description is obtained by directly


calculating the material derivative of the velocity in spatial description:
dv ( x, t ) ∂v ( x, t )
a ( x, t )
= = + v ( x, t ) ⋅∇v ( x, t )
dt ∂t
∂
∂  ωy   ∂x 
a ( x, t ) =   + [ ωy , −ωx ]  ∂  [ ωy , −ωx ]
∂t −ωx   
 ∂y 
∂ ∂ 
 ∂x ( ω y ) ( −ω x )  −ω2 x 
0  ∂x  0 −ω
  + [ ωy , −ωx ] =
∂ ∂
 [ ω
= y , −ω x ]    2 
0   ( ωy ) ( −ωx )  ω 0  −ω y 
 ∂y ∂y 

−ω x 
2

a ( x, t ) =  2 
−ω y 

47
1.6 Stationarity and Uniformity
Ch.1. Description of Motion

48
Stationary properties
 A property is stationary when its spatial description is not
dependent on time. REMARK
χ ( x,t ) = χ ( x ) In certain fields, the
term steady-state is
 The local derivative of a stationary property is zero. more commonly used.
∂χ ( x, t )
=χ ( x, t ) χ=
(x) 0
∂t

 The time-independence in the spatial description (stationarity) does


not imply time-independence in the material description:
=χ ( x, t ) χ=
(x) χ ( X, t ) χ ( X ) REMARK
This is easily understood if we consider,
for example, a stationary velocity:
v (= ( x ) v ( x ( X=
x, t ) v= , t ) ) V ( X, t )

49
Example
Consider a solid that rotates at a constant angular velocity ω and has the
following equation of motion:
=  x R sin (ωt + ϕ )


= y R cos (ωt + ϕ )

We have obtained:
Velocity in spatial description
 v x   ω y  stationary
v (=
x, t ) =   
 v y  −ω x 
Velocity in material description
Vx  − X ω sin ( ωt ) + Y ω cos ( ωt ) 
V (X
=, t ) =   
Vy   − X ω cos ( ω t ) − Y ω sin ( ω t ) 

50
Uniform properties
 A property is uniform when its spatial description is not
dependent on the spatial coordinates.
χ ( x, t ) = χ ( t )

If its spatial description does not depend on the coordinates (uniform



character of the property), neither does its material one.
=χ ( x, t ) χ=(t ) χ ( X, t ) χ ( t )

51
1.7 Trajectory (path-line)
Ch.1. Description of Motion

52
Trajectory or pathline
 A trajectory or pathline is the locus of the positions
occupied by a given particle in space throughout time.

REMARK
A trajectory can also be defined as
the path that a particle follows
through space as a function of time.

53
Equation of the trajectories
 The equation of a given particle’s trajectory is obtained
particularizing the equation of motion for that particle, which is
identified by it material coordinates X*.
 x ( t ) ϕ=( X , t ) X = X * φ (t )
= 

i (t )
 x= ϕ i ( X, t )= φi (t ) i ∈ 
X = X*

 Also, from the velocity field in spatial description, v(x,t):


 A family of curves is obtained from:
dx ( t )
= v=
dt
(x (t ) , t ) x φ ( C1 , C2 , C3 , t ) [1]

 Particularizing for a given particle by imposing the consistency condition in


the reference configuration:
= x ( t ) t = 0 X= X φ ( C1 , C2 , C3 , 0=
) Ci χ i ( X ) [2]
 Replacing [2] in [1], the equation of the trajectories in canonical form
( C1 ( X ) , C2 ( X ) , C3 ( X ) , t ) ϕ ( X, t )
x φ=
54
Example
Consider the following velocity field:
ωy 
v ( x, t ) =  
 −ω x 

Obtain the equation of the trajectories.

55
Example - Solution
We integrate the velocity field:
 dx ( t )
 = v x ( x, t )= ωy
dx ( t )
= v ( x, t )
dt

dt  dy ( t ) = v ( x, t ) = −ωx
 dt y

This is a crossed-variable system of differential equations. We derive the 2nd


eq. and replace it in the 1st one,

d 2 y (t ) dx ( t )
2
= −ω = −ω2
y (t ) y′′ + ω2 y = 0
dt dt

56
Example - Solution
The characteristic equation: r 2 + ω2 = 0
Has the characteristic solutions: rj =± i ω j ∈ {1, 2}
And the solution of the problem is:
(t ) Real Part {Z1eiwt + Z 2 e −=
y= iwt
} C1 cos ( ωt ) + C2 sin ( ωt )
And, using dy = −ωx , we obtain
dt
=− ( −C1ω sin ( ωt ) + C2 ω cos ( ωt ) )
1 dy 1
x =−
ω dt ω

So, the general solution is:


 x ( C1 , C=2,t) C1 sin ( ωt ) − C2 cos ( ωt )

 y ( C1 , C=
2,t) C1 cos ( ωt ) + C2 sin ( ωt )

57
 x ( C1 , C=2,t) C1 sin ( ωt ) − C2 cos ( ωt )
Example - Solution 
 y ( C1 , C=
2,t) C1 cos ( ωt ) + C2 sin ( ωt )

The canonical form is obtained from the initial conditions:


 = 0 =1 
 X x ( C ,=
=
 1 C2 , 0 ) C1 sin ( ω⋅ 0 ) − C2 cos =
( ω⋅ 0 ) C2
x ( C1 , C2 , 0 ) = X 
Y y C ,=
=
 ( 1 C2 , 0 ) C1 cos

( ω⋅0 ) + C2 sin=
( ω⋅ 0 ) 

C1
 = 1= 0

This results in:

= x Y sin ( ωt ) + X cos ( ωt )


x( X, t ) → 
=  y Y cos ( ωt ) − X sin ( ωt )

58
1.8 Streamline
Ch.1. Description of Motion

59
Streamline
 The streamlines are a family of curves which, for every
instant in time, are the velocity field envelopes.
time – t0 time – t1
Y Y
REMARK
Two streamlines can
never cut each other.
Is it true?
X X

 Streamlines are defined for any given time instant and change with
the velocity field.

REMARK
The envelopes of vector field are the curves whose tangent vector
at each point coincides (in direction and sense but not necessarily
in magnitude) with the corresponding vector of the vector field.

60
Equation of the Streamlines
 The equation of the streamlines is of
the type:
vz
d x d y dz dx
= = = d= λ ( ds ) = v vy
vx v y vz dλ
vx

 Also, from the velocity field in spatial description, v(x,t*) at a


given time instant t*:
 A family of curves is obtained from:
dx ( λ )

( x ( λ ) , t *)
v= x φ ( C1′, C2′ , C3′ , λ , t *)

 Where each group ( C1′, C2′ , C3′ ) identifies a streamline x(λ) whose
points are obtained assigning values to the parameter λ.
 For each time instant t* a new family of curves is obtained.

61
Trajectories and Streamlines
 For a stationary velocity field, the trajectories and the
streamlines coincide – PROOF:
1. If v(x,t)=v(x):
 Eq. trajectories:
dx ( t )
= v=
dt
( x (t ) , t ) x φ ( C1 , C2 , C3 , t )

 Eq. streamlines:
dx ( λ )
= v=

( x ( λ ) , t *) x φ ( C1 , C2 , C3 , λ , t *)

The differential equations only differ in the denomination of the


integration parameter (t or λ), so the solution to both systems MUST be
the same.

62
Trajectories and Streamlines
 For a stationary velocity field, the trajectories and the
streamlines coincide – PROOF:
2. If v(x,t)=v(x) the envelopes (i.e., the streamline) of the field do not
vary throughout time.

A particle’s trajectory is always tangent to the velocity field it


encounters at every time instant.

If a trajectory starts at a certain point in a streamline and the


streamline does not vary with time and neither does the velocity
field, the trajectory and streamline MUST coincide.

63
Trajectories and Streamlines
 The inverse is not necessarily true: if the trajectories and the
streamlines coincide, the velocity field is not necessarily
stationary – COUNTER-EXAMPLE:  at 
 Given the (non-stationary) velocity field:  v ( t )  =  0 
 0 

 The eq. trajectory are:


 a t 2 + C1 
 at   at   2 
 dx ( t )     0  dt
=  =0  dx ( t )  x(t) =  0 
   
 dt  0  0  0
   

 The eq. streamlines are:


 at   at   at λ + C1′ 
 dx ( λ )     0  dλ
=  =0  dx ( λ )    x ( t ) =  0 
 dλ   0   0   0 
 

64
Example
Consider the following velocity field:

xi
=vi i ∈ {1, 2,3}
1+ t
Obtain the equation of the trajectories and the streamlines associated to this
vector field.
Do they coincide? Why?

65
Example - Solution
 dx ( t )
 = v (x (t ) , t )
dt
Eq. trajectories: 
 dxi ( t ) v x ( t ) , t i ∈ 
= i ( )
 dt
Introducing the velocity field and rearranging:

dxi xi dxi dt
= i ∈  = i ∈ 
dt 1 + t xi 1 + t

Integrating both sides of the expression:


1 1
∫ xi i ∫ 1 + t dt
dx = ln xi= ln (1 + t ) + ln Ci= ln Ci (1 + t ) i ∈

The solution:
xi = Ci (1 + t ) i ∈
66
Example - Solution
 dx ( λ )
 = v ( x ( λ ) , t *)

Eq. streamlines:  dx t
 i ( ) v x ( λ ) , t * i ∈ 
= i ( )
 d λ
Introducing the velocity field and rearranging:

dxi xi dxi d λ
= i ∈  = i ∈ 
dλ 1+ t xi 1 + t
Integrating both sides of the expression:
  λ


+ Ki 
 λ 
 
=  1+ t Ki  1+ t 
 xi e= e e

1 1 λ
∫ xi i ∫ 1 + t d λ
dx = ln=
xi
1+ t
+ Ki
 i ∈  = Ci

The solution:  λ 
 
1+ t
=xi Ci e  
i ∈
67
Streamtube
A streamtube is a surface composed of streamlines
which pass through the points of a closed contour fixed
in space.

 In stationary cases, the tube will remain fixed in space


throughout time. In non-stationary cases, it will vary
(although the closed contour line is fixed).

69
1.9 Control and Material Surfaces
Ch.1. Description of Motion

81
Control Surface
 A control surface is a fixed surface in space which does not
vary in time.

=Σ: { x=
f ( x, y, z ) 0}

 Mass (particles) can flow across a control surface.

82
Material Surface
 A material surface is a mobile surface in the
space constituted always by the same particles.
 In the reference configuration, the surface Σ0 will be defined in terms
of the material coordinates:
=Σ0 : { X=
F ( X , Y , Z ) 0}
The set of particles (material points)
belonging the surface are the same at all times

 In spatial description F ( X ,=
Y , Z ) F ( X (x, t ), Y (x, t ), Z (=
x, t ) ) f=
(x, t ) f ( x, y, z , t )

=Σt : { x=
f ( x, y , z , t ) 0 }
 The set of spatial points belonging to the the surface depends on time
 The material surface moves in space
84
Material Surface
 Necessary and sufficient condition for a mobile surface in space,
implicitly defined by the function f ( x, y, z, t ) , to be a material
surface is that the material derivative of the function is zero:
 Necessary: if it is a material surface, its material description does not depend
on time:
d ∂F ( X)
f ( x, t ) → f ( x( X, t ), t ) =
F ( X, t ) 0 =f ( x, t ) = = 0
dt ∂t
 Sufficient: if the material derivative of f(x,t) is null:
d ∂F ( X, t )
f ( x, t ) → f ( x( X, t ), t ) =F ( X, t ) 0 = f ( x, t ) = F ( X, t ) ≡ F ( X)
dt ∂t
The surface Σt := { x f ( x, t ) = 0 } = { X F ( X ) = 0 }
contains always the same set the of particles (it is a material surface)

85
Control Volume
 A control volume is a group of fixed points in space
situated in the interior of a closed control surface, which
does not vary in time.
V := { x | f (x ) ≤ 0}

REMARK
The function f(x) is defined
so that f(x)<0 corresponds
to the points inside V.

 Particles can enter and exit a control volume.

87
Material Volume
 A material volume is a (mobile) volume enclosed inside a
material boundary or surface.
 In the reference configuration, the volume V0 will be defined in terms
of the material coordinates:
=V0 : { X | F ( X ) ≤ 0 }

 The particles X in the volume are


the same at all times

 In spatial description, the volume Vt will depend on time.


=Vt : {x | f ( x, t ) ≤ 0}

 The set of spatial points belonging to the the volume depends on time
 The material volume moves in space along time

89
Material Volume
 A material volume is always constituted by the same
particles. This is proved by reductio ad absurdum:

 If a particle is added into the volume, it would have to cross its


material boundary.

 Material boundaries are constituted always by the same particles, so,


no particles can cross.

 Thus, a material volume is always constituted by the same particles


(a material volume is a pack of particles).

90
Chapter 1
Description of Motion

rs
ee
s gin
1.1 Definition of the Continuous Medium

t d le En

r
A continuous medium is understood as an infinite set of particles (which form

ba
ge ro or
eS m
part of, for example, solids or fluids) that will be studied macroscopically, that

ci
f
is, without considering the possible discontinuities existing at microscopic level

ra
C d P cs
(atomic or molecular level). Accordingly, one admits that there are no discon-
b
a
i
tinuities between the particles and that the mathematical description of this
an an n

medium and its properties can be described by continuous functions.


y ha

le
liv or ec

1.2 Equations of Motion


M

.A

The most basic description of the motion of a continuous medium can be


m

achieved by means of mathematical functions that describe the position of each


d
uu

particle along time. In general, these functions and their derivatives are required
e
X Th

to be continuous.
er
tin
on

.O

Definition 1.1. Consider the following definitions:


C

• Spatial point: Fixed point in space.


©

• Material point: A particle. It may occupy different spatial points


during its motion along time.
• Configuration: Locus of the positions occupied in space by the
particles of the continuous medium at a given time t.

The continuous medium is assumed to be composed of an infinite number of


particles (material points) that occupy different positions in the physical space
during its motion along time (see Figure 1.1). The configuration of the contin-

1
2 C HAPTER 1. D ESCRIPTION OF M OTION

Ω0 – reference configuration
t0 – reference time
Ωt – present configuration
t – present time

rs
ee
Figure 1.1: Configurations of the continuous medium.

s gin
uous medium at time t, denoted by Ωt , is defined as the locus of the positions

t d le En
occupied in space by the material points (particles) of the continuous medium at

r
the given time.

ba
ge ro or
eS m
A certain time t = t0 of the time interval of interest is referred to as the ref-

ci
f
erence time and the configuration at this time, denoted by Ω0 , is referred to as

ra
C d P cs
initial, material or reference configuration1 .
b
a
i
Consider now the Cartesian coordinate system (X,Y, Z) in Figure 1.1 and the
an an n

corresponding orthonormal basis {ê1 , ê2 , ê3 }. In the reference configuration Ω0 ,


y ha

le
the position vector X of a particle occupying a point P in space (at the reference
liv or ec

time) is given by2,3


M

.A

X = X1 ê1 + X2 ê2 + X3 ê3 = Xi êi , (1.1)


m

where the components (X1 , X2 , X3 ) are referred to as material coordinates (of the
d
uu

particle) and can be collected in a vector of components denoted as4


e
X Th

er

⎡ ⎤
tin

X1
⎢ ⎥ de f
on

.O

not
X ≡ [X] = ⎣ X2 ⎦ = material coordinates. (1.2)
C

X3
©

1 In general, the time t0 = 0 will be taken as the reference time.


2 Notations (X,Y, Z) and (X1 , X2 , X3 ) will be used indistinctly to designate the Cartesian
coordinate system.
3 Einstein or repeated index notation will be used in the remainder of this text. Every repe-
tition of an index in the same monomial of an algebraic expression represents the sum over
that index. For example,
i=3 k=3 i=3 j=3
∑ Xi êi = Xi êi ∑ aik bk j = aik bk j ∑ ∑ ai j bi j = ai j bi j .
not not not
, and
i=1 k=1 i=1 j=1
4 Here, the vector (physical
not
entity) X is distinguished from its vector of components [X].
Henceforth, the symbol ≡ (equivalent notation) will be used to indicate that the tensor and
component notations at either side of the symbol are equivalent when the system of coordi-
nates used remains unchanged.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Equations of Motion 3

In the present configuration Ωt 5 , a particle originally located at a material


point P (see Figure 1.1) occupies a spatial point P and its position vector x is
given by
x = x1 ê1 + x2 ê2 + x3 ê3 = xi êi , (1.3)
where (x1 , x2 , x3 ) are referred to as spatial coordinates of the particle at time t,
⎡ ⎤
x1
not ⎢ ⎥ de f
x ≡ [x] = ⎣ x2 ⎦ = spatial coordinates. (1.4)
x3

rs
The motion of the particles of the continuous medium can now be described

ee
by the evolution of their spatial coordinates (or their position vector) along time.

s gin
Mathematically, this requires the definition of a function that provides for each
particle (identified by its label) its spatial coordinates xi (or its spatial position

t d le En
vector x) at successive instants of time. The material coordinates Xi of the par-

r
ticle can be chosen as the label that univocally characterizes it and, thus, the

ba
ge ro or
eS m
equation of motion

ci

f

ra
not
x = ϕ (particle,t) = ϕ (X,t) = x (X,t)
C d P cs
b
a
(1.5)
i
xi = ϕi (X1 , X2 , X3 ,t) i ∈ {1, 2, 3}
an an n
y ha

is obtained, which provides the spatial coordinates in terms of the material ones.
le
liv or ec

The spatial coordinates xi of the particle can also be chosen as label, defining
the inverse equation of motion6 as
M

.A

 not
X = ϕ −1 (x,t) = X (x,t) ,
m

(1.6)
d

Xi = ϕi−1 (x1 , x2 , x3 ,t) i ∈ {1, 2, 3} ,


uu
e
X Th

er
tin

which provides the material coordinates in terms of the spatial ones.


on

.O

Remark 1.1. There are different alternatives when choosing the la-
C

bel that characterizes a particle, even though the option of using its
material coordinates is the most common one. When the equation of
motion is written in terms of the material coordinates as label (as in
(1.5)), one refers to it as the equation of motion in canonical form.

5 Whenever possible, uppercase letters will be used to denote variables relating to the refer-
ence configuration Ω0 and lowercase letters to denote the variables referring to the current
configuration Ωt .
6 With certain abuse of notation, the function will be frequently confused with its image.
Hence, the equation of motion will be often written as x = x (X,t) and its inverse equation as
X = X (x,t).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
4 C HAPTER 1. D ESCRIPTION OF M OTION

There exist certain mathematical restrictions to guarantee the existence of ϕ


and ϕ −1 , as well as their correct physical meaning. These restrictions are:
• ϕ (X, 0) = X since, by definition, X is the position vector at the reference
time t = 0 (consistency condition).
• ϕ ∈ C1 (function ϕ is continuous with continuous derivatives at each point
and at each instant of time).
• ϕ is biunivocal (to guarantee that two particles do not occupy simultaneously
the same point in space and that a particle does not occupy simultaneously
more than one point in space).

∂ ϕ (X,t) not ∂ ϕ (X,t)

rs
• The Jacobian of the transformation J = det = > 0.
∂X ∂X

ee
s gin
The physical interpretation of this condition (which will be studied later) is
that every differential volume must always be positive or, using the principle of

t d le En
mass conservation (which will be seen later), the density of the particles must
always be positive.

r
ba
ge ro or
eS m
ci
f

ra
Remark 1.2. The equation of motion at the reference time t = 0 re-
C d P cs
b
a
sults in x (X,t)|t=0 = X. Accordingly, x = X, y = Y , z = Z is the
i
an an n

equation of motion at the reference time and the Jacobian at this in-
y ha

stant of time is7

∂ (xyz)
le
∂ xi
liv or ec


J (X, 0) = = det = det [δi j ] = det 1 = 1.
∂ (XY Z) ∂ Xj
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 1.2: Trajectory or pathline of a particle.

not
7The two-index operator Delta Kronecker = δi j is defined as δi j = 0 when i = j and δi j = 1
when i = j. Then, the unit tensor 1 is defined as [1]i j = δi j .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Equations of Motion 5

Remark 1.3. The expression x = ϕ (X,t), particularized for a fixed


value of the material coordinates X, provides the equation of the
trajectory or pathline of a particle (see Figure 1.2).

Example 1.1 – The spatial description of the motion of a continuous medium


is given by

rs
⎡ ⎤ ⎡ ⎤

ee
x1 = X1 e2t x = Xe2t
not ⎢ ⎥ ⎢ ⎥

s gin
x (X,t) ≡ ⎣ x2 = X2 e−2t ⎦ = ⎣ y = Y e−2t ⎦
x3 = 5X1t + X3 e2t z = 5Xt + Ze2t

t d le En

r
ba
Obtain the inverse equation of motion.

ge ro or
eS m
ci
f

ra
Solution
C d P cs
b
a
i
The determinant of the Jacobian is computed as
an an n


y ha


∂ x1 ∂ x1 ∂ x1
le
liv or ec



∂ X1 ∂ X2 ∂ X3 e2t 0 0
M

.A

∂ xi ∂ x
∂ x2 ∂ x2 = 0 e−2t 0 = e2t = 0.
J = =

2

m

∂ Xj ∂ X1 ∂ X2 ∂ X3
d

5t 0 e 2t
uu

∂ x3 ∂ x3 ∂ x3
e


X Th

∂X ∂ X2 ∂ X3
er
tin

1
on

.O

The sufficient (but not necessary) condition for the function x = ϕ (X,t) to
be biunivocal (that is, for its inverse to exist) is that the determinant of the
C

Jacobian of the function is not null. In addition, since the Jacobian is positive,
©

the motion has physical sense. Therefore, the inverse of the given spatial
description exists and is determined by
⎡ ⎤ ⎡ ⎤
X1 x1 e−2t
not ⎢ ⎥ ⎢ ⎥
X = ϕ −1 (x,t) ≡ ⎢ X
⎣ ⎦ ⎣
2
⎥=⎢ x 2 e 2t ⎥.

X3 x3 e−2t − 5tx1 e−4t

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
6 C HAPTER 1. D ESCRIPTION OF M OTION

1.3 Descriptions of Motion


The mathematical description of the properties of the particles of the continu-
ous medium can be addressed in two alternative ways: the material description
(typically used in solid mechanics) and the spatial description (typically used
in fluid mechanics). Both descriptions essentially differ in the type of argument
(material coordinates or spatial coordinates) that appears in the mathematical
functions that describe the properties of the continuous medium.

1.3.1 Material Description

rs
In the material description8 , a given property (for example, the density ρ) is
described by a certain function ρ (•,t) : R3 × R+ → R+ , where the argument (•)

ee
in ρ (•,t) represents the material coordinates,

s gin
ρ = ρ (X,t) = ρ (X1 , X2 , X3 ,t) . (1.7)

t d le En
Here, if the three arguments X ≡ (X1 , X2 , X3 ) are fixed, a specific particle is being

r
ba
ge ro or
eS m
followed (see Figure 1.3) and, hence, the name of material description.

ci
f

ra
C d P cs
1.3.2 Spatial Description
b
a
i
an an n

In the spatial description9 , the focus is on a point in space. The property is de-
y ha

scribed as a function ρ (•,t) : R3 × R+ → R+ of the point in space and of time,


le
liv or ec

ρ = ρ (x,t) = ρ (x1 , x2 , x3 ,t) . (1.8)


M

.A

Then, when the argument x in ρ = ρ (x,t) is assigned a certain value, the evolu-
m

tion of the density for the different particles that occupy the point in space along
d
uu

time is obtained (see Figure 1.3). Conversely, fixing the time argument in (1.8)
e
X Th

results in an instantaneous distribution (like a snapshot) of the property in space.


er
tin

Obviously, the direct and inverse equations of motion allow shifting from one
on

.O
C

Figure 1.3: Material description (left) and spatial description (right) of a property.

8 Literature on this topic also refers to the material description as Lagrangian description.
9 The spatial description is also referred to as Eulerian description.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Descriptions of Motion 7

description to the other as follows.



ρ (x,t) = ρ (x (X,t) ,t) = ρ (X,t)
(1.9)
ρ (X,t) = ρ (X (x,t) ,t) = ρ (x,t)

Example 1.2 – The equation of motion of a continuous medium is



x = X −Y t
not
x = x (X,t) ≡ y = Xt +Y .

rs
z = −Xt + Z

ee
Obtain the spatial description of the property whose material description is

s gin
X +Y + Z

t d le En
ρ (X,Y, Z,t) = .
1 + t2

r
ba
ge ro or
Solution

eS m
ci
f

ra
The equation of motion is given in the canonical form since in the reference
C d P cs
configuration Ω0 its expression results in
b
a
i
an an n


x=X
y ha

not
x = X (X, 0) ≡ y = Y .
le
liv or ec

z=Z
M

.A

The determinant of the Jacobian is


m

∂x ∂x ∂x
uu


1 −t 0
X Th

∂ X ∂Y ∂Z
er


tin

∂ xi ∂ y ∂ y
∂y
J = = = t 1 0 = 1 + t 2 = 0
∂ X j ∂ X ∂Y
on


.O

∂Z −t 0 1
∂z
C

∂z ∂z

©

∂ X ∂Y ∂Z
and the inverse equation of motion is given by
⎡ ⎤
x + yt
⎢X = ⎥
⎢ 1 + t2 ⎥
⎢ ⎥
not ⎢ y − xt ⎥
X (x,t) ≡ ⎢ Y = ⎥.
⎢ 1 + t 2 ⎥
⎢ ⎥
⎣ z + zt 2 + xt + yt 2 ⎦
Z=
1 + t2

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
8 C HAPTER 1. D ESCRIPTION OF M OTION

Consider now the material description of the property,


X +Y + Z
ρ (X,Y, Z,t) = ,
1 + t2
its spatial description is obtained by introducing the inverse equation of mo-
tion into the expression above,

x + yt + y + z + zt 2 + yt 2
ρ (X,Y, Z,t) ≡ = ρ (x, y, z,t) .
(1 + t 2 )2

rs
ee
s gin
1.4 Time Derivatives: Local, Material and Convective

t d le En
The consideration of different descriptions (material and spatial) of the proper-

r
ties of the continuous medium leads to diverse definitions of the time derivatives

ba
ge ro or
eS m
of these properties. Consider a certain property and its material and spatial de-

ci
scriptions,
f

ra
Γ (X,t) = γ (x,t) ,
C d P cs
(1.10)
b
a
i
an an n

in which the change from the spatial to the material description and vice versa
y ha

is performed by means of the equation of motion (1.5) and its inverse equa-
le
tion (1.6).
liv or ec
M

.A

Definition 1.2. The local derivative of a property is its variation


m

along time at a fixed point in space. If the spatial description γ (x,t)


d
uu
e

of the property is available, the local derivative is mathematically


X Th

er

written as10
tin

not ∂ γ (x,t)
local derivative = .
on

∂t
.O

The material derivative of a property is its variation along time fol-


C

lowing a specific particle (material point) of the continuous medium.


©

If the material description Γ (X,t) of the property is available, the


material derivative is mathematically written as

not d ∂Γ (X,t)
material derivative = Γ= .
dt ∂t

10 The expression ∂ (•,t)/∂t is understood in the classical sense of partial derivative with
respect to the variable t.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Time Derivatives: Local, Material and Convective 9

However, taking the spatial description of the property γ (x,t) and considering
the equation of motion is implicit in this expression yields

γ (x,t) = γ (x (X,t) ,t) = Γ (X,t) . (1.11)


Then, the material derivative (following a particle) is obtained from the spatial
description of the property as

not d ∂Γ (X,t)
material derivative = γ (x (X,t) ,t) = . (1.12)
dt ∂t

rs
Expanding (1.12) results in11

ee
dγ (x (X,t) ,t) ∂ γ (x,t) ∂ γ ∂ xi ∂ γ (x,t) ∂ γ ∂ x

s gin
= + = + · =
dt ∂t ∂ xi ∂t ∂t ∂ x 
∂t
(1.13)

t d le En
∂ γ (x,t) ∂ γ v (x,t)
= + · v (x,t) ,

r
∂t ∂x

ba
ge ro or
eS m
ci
f
where the definition of velocity as the derivative of the equation of motion (1.5)

ra
C d P cs
with respect to time has been taken into account,
b
a
i
an an n

∂ x (X,t)
y ha

= V (X (x,t) ,t) = v (x,t) . (1.14)


∂t
le
liv or ec

The deduction of the material derivative from the spatial description can be
M

.A

generalized for any property χ (x,t) (of scalar, vectorial or tensorial character)
as12
m

dχ (x,t) ∂ χ (x,t)
d
uu

= + v (x,t) · ∇χ (x,t) . (1.15)


e

dt ∂t
X Th

        
er
tin

material local convective


derivative derivative derivative
on

.O
C

Remark 1.4. The expression in (1.15) implicitly defines the convec-


tive derivative v · ∇ (•) as the difference between the material and
spatial derivatives of the property. In continuum mechanics, the term
convection is applied to phenomena that are related to mass (or par-
ticle) transport. Note that, if there is no convection (v = 0), the con-
vective derivative disappears and the local and material derivatives
coincide.

11 In literature, the notation D(•)/Dt is often used as an alternative to d(•)/dt.


12 The symbolic form of the spatial Nabla operator, ∇ ≡ ∂ êi /∂ xi , is considered here.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
10 C HAPTER 1. D ESCRIPTION OF M OTION

Example 1.3 – Given the equation of motion



x = X +Y t + Zt
not
x (X,t) ≡ y = Y + 2Zt ,
z = Z + 3Xt

and the spatial description of a property, ρ (x,t) = 3x + 2y + 3t, obtain the


material derivative of this property.

Solution

rs
The material description of the property is obtained introducing the equation

ee
of motion into its spatial description,

s gin
ρ (X,Y, Z,t) = 3 (X +Y t + Zt)+2 (Y + 2Zt)+3t = 3X +3Y t +7Zt +2Y +3t .

t d le En
The material derivative is then calculated as the derivative of the material

r
description with respect to time,

ba
ge ro or
eS m
ci
∂ρ
f

ra
= 3Y + 7Z + 3 .
C d P cs
∂t
b
a
i
an an n

An alternative way of deducing the material derivative is by using the concept


y ha

of material derivative of the spatial description of the property,


le
liv or ec

dρ ∂ρ
M

= + v · ∇ρ
.A

with
dt ∂t
m

∂ρ ∂x
d

=3, v= = [Y + Z, 2Z, 3X]T and ∇ρ = [3, 2, 0]T .


uu
e

∂t ∂t
X Th

er
tin

Replacing in the expression of the material derivative operator,


on

.O


= 3 + 3Y + 7Z
C

dt
©

is obtained. Note that the expressions for the material derivative obtained
from the material description, ∂ ρ/∂t, and the spatial description, dρ/dt, co-
incide.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Velocity and Acceleration 11

1.5 Velocity and Acceleration

Definition 1.3. The velocity is the time derivative of the equation of


motion.

The material description of velocity is, consequently, given by


rs

⎪ ∂ x (X,t)
⎨ V (X,t) =

ee
∂t (1.16)

s gin
⎪ ∂
⎩ Vi (X,t) = i (X,t) i ∈ {1, 2, 3}
x
∂t

t d le En
and, if the inverse equation of motion X = ϕ −1 (x,t) is known, the spatial de-

r
ba
ge ro or
scription of the velocity can be obtained as

eS m
ci
f

ra
v (x,t) = V (X (x,t) ,t) . (1.17)
C d P cs
b
a
i
an an n
y ha

Definition 1.4. The acceleration is the time derivative of the velocity


le
liv or ec

field.
M

.A
m

If the velocity is described in material form, the material description of the


uu
e

acceleration is given by
X Th

er


tin


⎪ ∂ V (X,t)
⎨ A (X,t) =
on

.O

∂t (1.18)
⎪ ∂V
⎩ Ai (X,t) = i (X,t) i ∈ {1, 2, 3}
C


©

∂t
and, through the inverse equation of motion X = ϕ −1 (x,t), the spatial descrip-
tion is obtained, a (x,t) = A (X (x,t) ,t). Alternatively, if the spatial description
of the velocity is available, applying (1.15) to obtain the material derivative of
v (x,t),
dv (x,t) ∂ v (x,t)
a (x,t) = = + v (x,t) · ∇v (x,t) , (1.19)
dt ∂t
directly yields the spatial description of the acceleration.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
12 C HAPTER 1. D ESCRIPTION OF M OTION

Example 1.4 – Consider the solid in the figure below, which rotates at a
constant angular velocity ω and has the expression

x = R sin (ωt + φ )
y = R cos (ωt + φ )

as its equation of motion. Find the velocity and acceleration of the motion
described both in material and spatial forms.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
Solution
f

ra
C d P cs
b
a
The equation of motion can be rewritten as
i
an an n


y ha

x = R sin (ωt + φ ) = R sin (ωt) cos φ + R cos (ωt) sin φ


le
y = R cos (ωt + φ ) = R cos (ωt) cos φ − R sin (ωt) sin φ
liv or ec
M

.A

and, since for t = 0, X = R sin φ and Y = R cos φ , the canonical form of the
m

equation of motion and its inverse equation result in


d

 
uu
e

x = X cos (ωt) +Y sin (ωt) X = x cos (ωt) − y sin (ωt)


X Th

er

.
tin

and
y = −X sin (ωt) +Y cos (ωt) Y = x sin (ωt) + y cos (ωt)
on

.O

Velocity in material description:


C

⎡ ⎤
©

∂x
∂ x (X,t) not ⎢ = −Xω sin (ωt) +Y ω cos (ωt) ⎥
V (X,t) = )≡⎢ ∂t ⎥
∂t ⎣ ⎦
∂y
= −Xω cos (ωt) −Y ω sin (ωt)
∂t

Velocity in spatial description:


Replacing the canonical form of the equation of motion into the material
description of the velocity results in

not ωy
v (x,t) = V (X (x,t) ,t) ≡ .
−ωx

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Velocity and Acceleration 13

Acceleration in material description:


∂ V (X,t)
A (X,t) =
∂t
⎡ ⎤
∂ vx
⎢ = −Xω 2 cos (ωt) −Y ω 2 sin (ωt) ⎥
not ⎢ ∂t ⎥
A (X,t) ≡ ⎢ ⎥=
⎣ ∂v ⎦
y
= Xω 2 sin (ωt) −Y ω 2 cos (ωt)
∂t

rs
X cos (ωt) +Y sin (ωt)

ee
= −ω 2
−X sin (ωt) +Y cos (ωt)

s gin
Acceleration in spatial description:

t d le En
Replacing the canonical form of the equation of motion into the material

r
description of the acceleration results in

ba
ge ro or
eS m

ci
f −ω 2 x

ra
not
a (x,t) = A (X (x,t) ,t) ≡ .
C d P cs
b
a
−ω 2 y
i
an an n
y ha

This same expression can be obtained if the expression for the velocity v (x,t)
le
liv or ec

and the definition of material derivative in (1.15) are taken into account,
M

.A

dv (x,t) ∂ v (x,t)
a (x,t) = = + v (x,t) · ∇v (x,t) =
∂t
m

dt ⎡ ⎤
d
uu



e

∂  ⎢ ⎥
⎢ ∂x ⎥ 
X Th

ωy
er

not
tin

≡ + ωy , −ωx ⎢ ⎥ ωy , −ωx ,
∂t −ωx ⎣ ∂ ⎦
on

.O

∂y
⎡ ⎤
C

∂ ∂
  ⎢ (ωy) (−ωx) ⎥ −ω 2x
0 ⎢ ⎥
+ ωy , −ωx ⎢ ∂ x ∂x
not
a (x,t) ≡ ⎥= .
0 ⎣ ∂ ∂ ⎦ −ω 2 y
(ωy) (−ωx)
∂y ∂y
Note that the result obtained using both procedures is identical.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
14 C HAPTER 1. D ESCRIPTION OF M OTION

1.6 Stationarity

Definition 1.5. A property is stationary when its spatial description


does not depend on time.

According to the above definition, and considering the concept of local deriva-
tive, any stationary property has a null local derivative. For example, if the ve-

rs
locity for a certain motion is stationary, it can be described in spatial form as

ee
∂ v (x,t)

s gin
v (x,t) = v (x) ⇐⇒ =0. (1.20)
∂t

t d le En

r
ba
Remark 1.5. The non-dependence on time of the spatial description

ge ro or
eS m
ci
(stationarity) assumes that, for a same point in space, the property
f

ra
being considered does not vary along time. This does not imply that,
C d P cs
b
a
for a same particle, such property does not vary along time (the ma-
i
an an n

terial description may depend on time). For example, if the velocity


y ha

v (x,t) is stationary,
le
v (x,t) ≡ v (x) = v (x (X,t)) = V (X,t) ,
liv or ec
M

.A

and, thus, the material description of the velocity depends on time.


In the case of stationary density (see Figure 1.4), for two particles
m

labeled X1 and X2 that have varying densities along time, when oc-
uu
e

cupying a same spatial point x (at two different times t1 and t2 ) their
X Th

er
tin

density value will coincide,


ρ (X1 ,t1 ) = ρ (X2 ,t2 ) = ρ (x) .
on

.O
C

That is, for an observer placed outside the medium, the density of
©

the fixed point in space x will always be the same.

Figure 1.4: Motion of two particles with stationary density.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Trajectory 15

Example 1.5 – Justify if the motion described in Example 1.4 is stationary


or not.

Solution
The velocity field in Example 1.4 is v (x) ≡ [ωy , −ωx]T . Therefore, it is a
not

case in which the spatial description of the velocity is not dependent on time
and, thus, the velocity is stationary. Obviously, this implies that the velocity
of the particles (whose motion is a uniform rotation with respect to the origin,
with angular velocity ω) does not depend on time (see figure below). The

rs
direction of the velocity vector for a same particle is tangent to its circular
trajectory and changes along time.

ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

The acceleration (material derivative of the velocity),


le
liv or ec

dv (x) ∂ v (x)
a (x) = = + v (x) · ∇v (x) = v (x) · ∇v (x) ,
M

.A

dt ∂t
m

appears due to the change in direction of the velocity vector of the particles
d
uu

and is known as the centripetal acceleration.


e
X Th

er
tin

1.7 Trajectory
on

.O
C

Definition 1.6. A trajectory (or pathline) is the locus of the positions


occupied in space by a given particle along time.

The parametric equation of a trajectory as a function of time is obtained by par-


ticularizing the equation of motion for a given particle (identified by its material
coordinates X∗ , see Figure 1.5),


x (t) = ϕ (X,t) ∗
. (1.21)
X=X

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
16 C HAPTER 1. D ESCRIPTION OF M OTION

Figure 1.5: Trajectory or pathline of a particle.

Given the equation of motion x = ϕ (X,t), each point in space is occupied

rs
by a trajectory characterized by the value of the label (material coordinates) X.
Then, the equation of motion defines a family of curves whose elements are the

ee
trajectories of the various particles.

s gin
1.7.1 Differential Equation of the Trajectories

t d le En
Given the velocity field in spatial description v (x,t), the family of trajectories

r
ba
ge ro or
can be obtained by formulating the system of differential equations that imposes

eS m
ci
that, for each point in space x, the velocity vector is the time derivative of the
f

ra
parametric equation of the trajectory defined in (1.21), i.e.,
C d P cs
b
a

i
⎪ dx (t)
an an n


⎨ = v (x (t) , t) ,
y ha

Find x (t) := dt (1.22)



le
⎩ dxi (t) = vi (x (t) , t) i ∈ {1, 2, 3} .

liv or ec

dt
M

.A

The solution to this first-order system of differential equations depends on three


m

integration constants (C1 ,C2 ,C3 ),


uu


e

x = φ (C1 ,C2 ,C3 , t) ,


X Th

er
tin

(1.23)
x = φi (C1 ,C2 ,C3 , t) i ∈ {1, 2, 3} .
on

.O

These expressions constitute a family of curves in space parametrized by the


C

constants (C1 ,C2 ,C3 ). Assigning a particular value to these constants yields a
©

member of the family, which is the trajectory of a particle characterized by the


label (C1 ,C2 ,C3 ).
To obtain the equation in canonical form, the consistency condition is im-
posed in the reference configuration,


x (t) = X =⇒ X = φ (C1 ,C2 ,C3 , 0) =⇒ Ci = χi (X) i ∈ {1, 2, 3} , (1.24)
t=0

and, replacing into (1.23), the canonical form of the equation of the trajectory,
X = φ (C1 (X) ,C2 (X) ,C3 (X) , t) = ϕ (X,t) , (1.25)
is obtained.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Trajectory 17

Example 1.6 – Given the velocity field in Example 1.5, v (x) ≡ [ωy , −ωx]T ,
not

obtain the equation of the trajectory.

Solution
Using expression (1.22), one can write


⎪ dx (t)
dx (t) ⎨ = vx (x,t) = ωy ,
= v (x,t) =⇒ dt

⎩ dy (t) = vy (x,t) = −ωx .
dt ⎪

rs
dt

ee
s gin
This system of equations is a system with crossed variables. Differentiating
the second equation and replacing the result obtained into the first equation

t d le En
yields
d 2 y (t) dx (t)
= −ω = −ω 2 y (t) =⇒ y + ω 2 y = 0 .

r
ba
ge ro or
dt 2 dt

eS m
ci
f
The characteristic equation of this second-order differential equation is

ra
C d P cs
r2 + ω 2 = 0 and its characteristic solutions are r j = ±iω j ∈ {1, 2}.
b
a
i
Therefore, the y component of the equation of the trajectory is
an an n
y ha

 
y (t) = Real Part C1 eiwt +C2 e−iwt = C1 cos (ωt) +C2 sin (ωt) .
le
liv or ec

The solution for x (t) is obtained from dy/dt = −ωx , which results in
M

.A

x = −dy/(ω dt) and, therefore,



m

x (C1 ,C2 ,t) = C1 sin (ωt) −C2 cos (ωt) ,


uu
e

y (C1 ,C2 ,t) = C1 cos (ωt) +C2 sin (ωt) .


X Th

er
tin

This equation provides the expressions of the trajectories in a non-canonical


on

.O

form. The canonical form is obtained considering the initial condition,


C

x (C1 ,C2 , 0) = X ,
©

that is, 
x (C1 ,C2 , 0) = −C2 = X ,
y (C1 ,C2 , 0) = C1 = Y .
Finally, the equation of motion, or the equation of the trajectory, in canonical
form 
x = Y sin (ωt) + X cos (ωt)
y = Y cos (ωt) − X sin (ωt)
is obtained.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
18 C HAPTER 1. D ESCRIPTION OF M OTION

1.8 Streamline

Definition 1.7. The streamlines are a family of curves that, for every
instant of time, are the velocity field envelopes13 .

rs
According to its definition, the tangent at each point of a streamline has the same

ee
direction (though not necessarily the same magnitude) as the velocity vector at

s gin
that same point in space.

t d le En

r
Remark 1.6. In general, the velocity field (in spatial description) will

ba
ge ro or
eS m
be different for each instant of time (v ≡ v (x,t)). Therefore, one

ci
f
must speak of a different family of streamlines for each instant of

ra
C d P cs
time (see Figure 1.6).
b
a
i
an an n
y ha

le
liv or ec

1.8.1 Differential Equation of the Streamlines


M

.A

Consider a given time t ∗ and the spatial description of the velocity field at this
m

time v (x,t ∗ ). Let x (λ ) be the equation of a streamline parametrized in terms of


d
uu

a certain parameter λ . Then, the vector tangent to the streamline is defined, for
e
X Th

er
tin
on

.O
C

Figure 1.6: Streamlines at two different instants of time.

13The envelopes of a vector field are the family of curves whose tangent vector has, at each
point, the same direction as the corresponding vector of the vector field.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Streamline 19

each value of λ 14 , by dx (λ )/dλ and the vector field tangency condition can be
written as follows.


⎪ dx (λ )
⎨ = v (x (λ ) , t ∗ ) ,
Find x (λ ) := dλ (1.26)

⎩ dxi (λ ) = vi (x (λ ) , t ∗ ) i ∈ {1, 2, 3} .


The expressions in (1.26) constitute a system of first-order differential equa-
tions whose solution for each time t ∗ , which will depend on three integration

rs
constants (C1 ,C2 ,C3 ), provides the parametric expression of the streamlines,

ee

x = φ (C1 ,C2 ,C3 , λ , t ∗ ) ,

s gin
(1.27)
xi = φi (C1 ,C2 ,C3 , λ , t ∗ ) i ∈ {1, 2, 3} .

t d le En
Each triplet of integration constants (C1 ,C2 ,C3 ) identifies a streamline whose

r
ba
ge ro or
points, in turn, are obtained by assigning values to the parameter λ . For each

eS m
ci
time t ∗ a new family of streamlines is obtained.
f

ra
C d P cs
b
a
i
an an n

Remark 1.7. In a stationary velocity field (v (x,t) ≡ v (x)) the trajec-


y ha

tories and streamlines coincide. This can be proven from two differ-
le
liv or ec

ent viewpoints:
M

.A

• The fact that the time variable does not appear in (1.22) or (1.26)
m

means that the differential equations defining the trajectories and


d

those defining the streamlines only differ in the denomination of


uu
e

the integration parameter (t or λ , respectively). The solution to


X Th

er
tin

both systems must be, therefore, the same, except for the name
of the parameter used in each type of curves.
on

.O

• From a more physical point of view: a) If the velocity field is


C

stationary, its envelopes (the streamlines) do not change along


©

time; b) a given particle moves in space keeping the trajectory


in the direction tangent to the velocity field it encounters along
time; c) consequently, if a trajectory starts at a certain point in a
streamline, it will stay on this streamline throughout time.

14 It is assumed that the value of the parameter λ is chosen such that, at each point in space
x, not only does dx (λ )/dλ have the same direction as the vector v (x,t), but it coincides
therewith.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
20 C HAPTER 1. D ESCRIPTION OF M OTION

1.9 Streamtubes

Definition 1.8. A streamtube is a surface formed by a bundle of


streamlines that occupy the points of a closed line, fixed in space,
and that does not constitute a streamline.

In non-stationary cases, even though the closed line does not vary in space, the

rs
streamtube and streamlines do change. On the contrary, in a stationary case, the

ee
streamtube remains fixed in space along time.

s gin
1.9.1 Equation of the Streamtube

t d le En
Streamlines constitute a family of curves of the type

r
ba
x = f (C1 ,C2 ,C3 , λ , t) .

ge ro or
(1.28)

eS m
ci
f

ra
The problem consists in determining, for each instant of time, which curves
C d P cs
b
a
of the family of curves of the streamlines cross a closed line, which is fixed in the
i
space Γ , whose mathematical expression parametrized in terms of a parameter s
an an n
y ha

is
Γ := x = g (s) .
le
(1.29)
liv or ec

To this aim, one imposes, in terms of the parameters λ ∗ and s∗ , that a same point
M

.A

belong to both curves,


m


d

g (s∗ ) = f (C1 ,C2 ,C3 , λ ∗ , t) ,


uu
e

(1.30)
X Th

gi (s∗ ) = fi (C1 ,C2 ,C3 , λ ∗ , t) i ∈ {1, 2, 3} .


er
tin
on

A system of three equations is obtained from which, for example, s∗ , λ ∗ and C3


.O

can be isolated,
C

s∗ = s∗ (C1 ,C2 , t) ,
©

λ ∗ = λ ∗ (C1 ,C2 , t) , (1.31)


C3 = C3 (C1 ,C2 , t) .
Introducing (1.31) into (1.30) yields

x = f (C1 , C2 , C3 (C1 ,C2 , t) , λ ∗ (C1 ,C2 , t) , t) = h (C1 ,C2 , t) , (1.32)


which constitutes the parametrized expression (in terms of the parameters C1
and C2 ) of the streamtube for each time t (see Figure 1.7).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Streaklines 21

Figure 1.7: Streamtube at a given time t.

rs
ee
s gin
1.10 Streaklines

t d le En

r

ba
ge ro or
Definition 1.9. A streakline, relative to a fixed point in space x

eS m
named spill point and at a time interval ti ,t f named spill period,

ci
f

ra
is the locus of the positions occupied at time  t by all the particles
C d P cs
b
a
that have occupied x∗ over the time τ ∈ [ti ,t] ti ,t f .
i
an an n
y ha

le
liv or ec

The above definition corresponds to the physical concept of the color line
M

.A

(streak) that would be observed in the medium at time  t if a tracer fluid were
injected at spill point x∗ throughout the time interval ti ,t f (see Figure 1.8).
m

d
uu
e
X Th

er
tin
on

.O
C

 
Figure 1.8: Streakline corresponding to the spill period τ ∈ ti ,t f .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
22 C HAPTER 1. D ESCRIPTION OF M OTION

1.10.1 Equation of the Streakline


To determine the equation of a streakline one must identify all the particles that
occupy point x∗ in the corresponding times τ. Given the equation of motion (1.5)
and its inverse equation (1.6), the label of the particle which at time τ occupies
the spill point must be identified. Then,

x∗ = x (X, τ)
=⇒ X = f (τ) (1.33)
xi∗ = xi (X, τ) i ∈ {1, 2, 3}

rs
and replacing (1.33) into the equation of motion (1.5) results in

ee
 
x = ϕ (f (τ) , t) = g (τ, t) τ ∈ [ti ,t] ti ,t f . (1.34)

s gin
Expression (1.34) is, for each time t, the parametric expression (in terms of

t d le En
parameter τ) of a curvilinear segment in space which is the streakline at that

r
time.

ba
ge ro or
eS m
ci
f

ra
C d P cs
Example 1.7 – Given the equation of motion
b
a

i
an an n

⎨ x = (X +Y )t 2 + X cost ,
y ha

⎩ y = (X +Y ) cost − X ,
le
liv or ec
M

.A

obtain the equation of the streakline associated with the spill point x∗ = (0, 1)
for the spill period [t0 , +∞).
m

d
uu
e

Solution
X Th

er
tin

The material coordinates of a particle that has occupied the spill point at time
on

τ are given by
.O


 ⎪ −τ 2
C


⎨ X = ,
0 = (X +Y ) τ + X cos τ
2
©

=⇒ τ 2 + cos2 τ
1 = (X +Y ) cos τ − X ⎪
⎩ Y = τ + cos τ .
⎪ 2

τ 2 + cos2 τ
Therefore, the label of the particles that have occupied the spill point from
the initial spill time t0 until the present time t is defined by

−τ 2 ⎪

X= 2 ⎬ 
τ + cos τ 2
τ ∈ [t0 ,t] [t0 , ∞) = [t0 ,t] .
τ 2 + cos τ ⎪⎪

Y= 2
τ + cos2 τ

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Material Surface 23

Then, replacing these into the equation of motion, the equation of the streak-
line is obtained,
⎡ ⎤
cos τ −τ 2
not ⎢
x= 2 t + 2
2 cost ⎥
x = g (τ,t) ≡ ⎢ τ + cos2 τ τ + cos2 τ ⎥ τ ∈ [t0 ,t] .
⎣ ⎦
cos τ −τ 2
y= 2 cost − 2
τ + cos2 τ τ + cos2 τ

rs
Remark 1.8. In a stationary problem, the streaklines are segments of

ee
the trajectories (or of the streamlines). The rationale is based on the

s gin
fact that, in the stationary case, the trajectory follows the envelope of
the velocity field, which remains constant along time. If one consid-
ers a spill point x∗ , all the particles that occupy this point will follow

t d le En
portions (segments) of the same trajectory.

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
1.11 Material Surface
i
an an n
y ha

le
liv or ec

Definition 1.10. A material surface is a mobile surface in space al-


M

.A

ways constituted by the same particles (material points).


m

d
uu
e
X Th

In the reference configuration Ω0 , surface Σ0 can be defined in terms of a func-


er
tin

tion of the material coordinates F (X,Y, Z) as


on

.O

Σ0 := {X,Y, Z | F (X,Y, Z) = 0} . (1.35)


C

Remark 1.9. The function F (X,Y, Z) does not depend on time,


which guarantees that the particles, identified by their label, that sat-
isfy equation F (X,Y, Z) = 0 are always the same in accordance with
the definition of material surface.

The spatial description of the surface is obtained from the spatial description
of F (X (x,t)) = f (x, y, z,t) as

Σt := {x, y, z | f (x, y, z,t) = 0} . (1.36)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
24 C HAPTER 1. D ESCRIPTION OF M OTION

Remark 1.10. The function f (x, y, z,t) depends explicitly on time,


which indicates that the points in space that are on the surface will
vary along time. This time dependence of the spatial description of
the surface confers the character of mobile surface in space to the
surface (see Figure 1.9).

rs
ee
Remark 1.11. The necessary and sufficient condition for a mobile
surface in space, defined implicitly by a function f (x, y, z,t) = 0, to

s gin
be material (to be always constituted by the same particles) is that
the material derivative of f (x, y, z,t) is null,

t d le En

r
d f (x,t) ∂ f

ba
ge ro or
= +v·∇f = 0 ∀x ∈ Σt ∀t .

eS m
∂t

ci
dt
f

ra
C d P cs
The condition is necessary because, if the surface is a material sur-
b
a
face, its material description will not depend on time (F ≡ F (X))
i
an an n

and, therefore, its spatial description will have a null material deriva-
y ha

tive. The condition of sufficiency is based on the fact that, if the ma-
le
liv or ec

terial derivative of f (x,t) is zero, the corresponding material de-


scription will not depend on time (F ≡ F (X)) and, therefore, the set
M

.A

of particles (identified by their material coordinates) that satisfy the


m

condition F (X) = 0 is always the same.


d
uu
e
X Th

er
tin
on

.O
C

Figure 1.9: A material surface at two different instants of time.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Material Surface 25

Example 1.8 – In ocean waves theory, the condition that the free surface
of the fluid in contact with the atmosphere is a material surface is imposed.
This restriction implies that the free surface is always composed of the same
particles, which is a reasonable hypothesis (especially in deep waters). De-
termine how this condition is stated in terms of the velocity field of the fluid.

Solution
Assuming that z = η (x, y,t) defines the elevation of the sea surface with re-

rs
spect to a reference level, the free surface of the water will be given by

ee
f (x, y, z,t) ≡ z − η (x, y,t) = 0 .

s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A

The condition d f /dt = 0 can be written as


m

∂f ∂f ∂η
uu

df
e

= + v · ∇ f where =− and
X Th

∂t ∂t ∂t
er

dt
tin

∂f ∂f ∂f T ∂f ∂f ∂f
on

.O

not
v · ∇ f ≡ [vx , vy , vz ] , , = vx + vy + vz .
∂x ∂y ∂z ∂x ∂y ∂z
C

Then,
df ∂f ∂η ∂η ∂η
= +v·∇f = − − vx − vy + vz = 0
dt ∂t ∂t ∂x ∂y
and, isolating vz leads to
∂η ∂η ∂η
vz = + vx + vy .
∂t ∂x ∂y
Therefore, the material surface condition results in a condition on the vertical
component of the velocity field.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
26 C HAPTER 1. D ESCRIPTION OF M OTION

1.12 Control Surface

Definition 1.11. A control surface is a fixed surface in space.

The mathematical description of a control surface is given by


Σ := {x | f (x, y, z) = 0} . (1.37)

rs
Obviously, a control surface is occupied by the different particles of the contin-

ee
uous medium along time (see Figure 1.10).

s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec

Figure 1.10: Movement of particles through a control surface along time.


M

.A
m

1.13 Material Volume


d
uu
e
X Th

er
tin

Definition 1.12. A material volume is a volume enclosed by a closed


on

.O

material surface.
C

The mathematical description of a material volume (see Figure 1.11) is given, in


the material description, by15
V0 := {X | F (X) ≤ 0} (1.38)

and, in the spatial description, by


Vt := {x | f (x,t) ≤ 0} , (1.39)
15 It is assumed that function F (X) is defined such that F (X) < 0 corresponds to points in
the interior of V0 .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Control Volume 27

where F (X) = f (x (X,t) ,t) is the function that describes the material surface
that encloses the volume.

Remark 1.12. A material volume is always constituted by the same


particles. This is proven by reductio ad absurdum as follows. If a
certain particle could enter or exit the material volume, it would be
incorporated into the material surface during its motion (at least, for
an instant of time). This would be contrary to the fact that the sur-
face, being a material surface, is always constituted by the same par-

rs
ticles.

ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A

Figure 1.11: A material volume at two different instants of time.


m

d
uu
e

1.14 Control Volume


X Th

er
tin
on

.O

Definition 1.13. A control volume is a group of points in space situ-


C

ated in the interior of a closed control surface.


©

It is a volume fixed in space that is occupied by the particles of the medium


during its motion. The mathematical description of the control volume (see Fig-
ure 1.12) is16
V := {x | f (x) ≤ 0} . (1.40)

16 It is assumed that function f (x) is defined such that f (x) < 0 corresponds to points in the
interior of V .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
28 C HAPTER 1. D ESCRIPTION OF M OTION

rs
Figure 1.12: A control volume is occupied by different particles along time.

ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 29

P ROBLEMS

Problem 1.1 – Justify whether the following statements are true or false.
a) If the velocity field is stationary, the acceleration field is also stationary.
b) If the velocity field is uniform, the acceleration field is always null.

rs
ee
Solution

s gin
a) A stationary velocity field implies that the spatial description of velocity does

t d le En
not depend on time,
∂ v (x,t)
= 0 =⇒ v (x) .

r
ba
∂t

ge ro or
eS m
ci
The acceleration is the material derivative of the velocity, therefore
f

ra
C d P cs
b
a
∂ v (x,t)
i
a (x,t) = + v (x,t) · ∇v (x,t) = v (x) · ∇v (x) .
an an n

∂t
y ha

le
The resulting expression does not depend on time. Thus, the statement is true.
liv or ec
M

.A

b) A uniform velocity field implies that the spatial description of velocity does
not depend on the spatial coordinates,
m

d
uu

v (x,t) =⇒ v (t) .
e
X Th

er
tin

The material derivative of the velocity results in


on

.O

∂ v (x,t) ∂ v (t)
a (x,t) = + v (x,t) · ∇v (x,t) = ,
C

∂t ∂t
©

where the expression used for the gradient of the velocity field is
∂ vi (t)
[∇v (t)]i j = =0.
∂xj

Therefore, the statement is false because ∂ v (t)/∂t is not necessarily zero.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
30 C HAPTER 1. D ESCRIPTION OF M OTION

Problem 1.2 – Calculate the acceleration at time t = 2 in point (1, 1, 1) of the


velocity field
not  ! T
v ≡ x − z , z e t + e−t , 0 .

Solution
Since the velocity field is given in its spatial expression and the acceleration is
requested for a point x∗ = (1, 1, 1)T , the equation of motion is not needed. One

rs
can simply apply

ee
dv (x,t) ∂ v (x,t)

s gin
a (x,t) = = + v (x,t) · ∇v (x,t) ,
dt ∂t

t d le En
where
∂ v not  ! T

r
≡ 0, z e t − e−t , 0 and

ba
ge ro or
∂t

eS m
⎡ ⎤

ci
f

ra

C d P cs
⎢ ⎥ ⎡ ⎤
b
a
⎢ ∂x ⎥
i
⎢ ⎥ 1 0 0
an an n

not ⎢ ∂ ⎥  !  ⎢ ⎥
y ha

∇v ≡ ⎢ ⎢ ⎥ x−z , z e t + e−t , 0 =⎢ 0⎥
⎥ ⎣ 0 0 ⎦ , such that
le
⎢ ∂y ⎥
liv or ec

⎢ ⎥ −1 (e t + e−t ) 0
⎣ ∂ ⎦
M

.A

∂z
m

0]T .
not
v · ∇v ≡ [x − z , 0,
uu
e
X Th

er
tin

Therefore, the spatial expression for the acceleration field is


on

.O

not  ! T
a ≡ x − z , z e t − e−t , 0
C

and, for the given point at the given instant of time, the acceleration is

not  T
a (x = x∗ , t = 2) ≡ 0 , e2 − e−2 , 0 .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 31

Problem 1.3 – The equation of a certain motion is


1 ! 1 !
x=X , y= (Y + Z) e t + (Y − Z) e−t , z= (Y + Z) e t − (Y − Z) e−t .
2 2
Calculate the accelerations that would be observed along time by:
a) An observer located in the fixed point (1, 1, 1).
b) An observer traveling with the particle that at time t = 0 occupied position
(1, 1, 1).

rs
c) An observer located in point (1, 1, 1) that measures the accelerations as the

ee
difference between velocities at this point per unit of time.

s gin
t d le En
Solution

r
ba
a) The spatial description of the acceleration in point x∗ = (1, 1, 1) must be

ge ro or
eS m
ci
obtained,
f

ra
∂ V (X (x∗ ,t) ,t)
C d P cs
a (x = x∗ , t) = A (X (x∗ ,t) , t) = .
b
a
∂t
i
an an n

The material expression of the velocity field is


y ha

⎡ ⎤
le
liv or ec

⎢ 0 ⎥
M

.A

⎢ ⎥
∂ x (X,t) ⎢ ⎥
not ⎢ ⎥
V (X,t) ≡ ⎢ 1 ((Y + Z) e t − (Y − Z) e−t ) ⎥ .
m

V (X,t) = =⇒
d

∂t ⎢2 ⎥
uu

⎢ ⎥
e

⎣1 ⎦
X Th

er

−t
((Y + Z) e + (Y − Z) e )
t
tin

2
on

.O

Then, the material description of the acceleration is


C

⎡ ⎤
©

⎢ 0 ⎥
⎢ ⎥
∂ V (X,t) not ⎢
⎢1


A (X,t) = ≡ ⎢ ((Y + Z) e t + (Y − Z) e−t ) ⎥ .
∂t ⎢2 ⎥
⎢ ⎥
⎣1 ⎦
((Y + Z) e t − (Y − Z) e−t )
2
Careful observation of the expression obtained reveals that
1 !
Ay = (Y + Z) e t + (Y − Z) e−t = y and
2

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
32 C HAPTER 1. D ESCRIPTION OF M OTION

1 !
(Y + Z) e t − (Y − Z) e−t = z .
Az =
2
Therefore, the spatial description of the acceleration field is

z]T
not
a (x,t) ≡ [0 , y,

and, for x = x∗ ,
a (x∗ ,t) ≡ [0 , 1 , 1]T .
not

rs
NOTE: In case one does not realize that Ay = y and Az = z, this same result can

ee
be obtained by replacing into the material expression of the acceleration field
the inverse equation of motion as follows.

s gin
 
y + z = (Y + Z) e t Y + Z = (y + z) e−t

t d le En
=⇒
y − z = (Y − Z) e−t Y − Z = (y − z) e t

r
ba
ge ro or
eS m

ci

⎪ f

ra
⎪ X =x
C d P cs


b
a


i
an an n

1
Y = ((y + z) e−t + (y − z) e t )
y ha


⎪ 2


le

⎪ 1
liv or ec

⎩ Z = ((y + z) e−t − (y − z) e t )
2
M

.A
m

b) The material description of the acceleration in point X∗ = (1, 1, 1) must be


uu
e

obtained. Replacing point X∗ into the expression obtained in a) yields


X Th

er
tin

not  T
A (X∗ ,t) ≡ 0 , e t , e t .
on

.O
C

c) The difference between the spatial velocities per unit of time must be ob-
tained, for point x∗ = (1, 1, 1),

Δ v (x∗ ,t) ∂ v (x∗ ,t)


−→ .
Δt ∂t
The spatial description of the velocity field is

v (x = x∗ , t) = V (X (x∗ ,t) ,t) .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 33

Careful observation of the material expression of the velocity field obtained in a)


reveals that Vy = z and Vz = y, therefore

∂ v (x∗ ,t) not


v (x,t) ≡ [0 , z , y]T ≡ [0 , 0 , 0]T .
not
=⇒
∂t

Problem 1.4 – Given the spatial description of the velocity field in Cartesian
coordinates,

rs
v ≡ [x , y , zϕ (t)]T
not

ee
and the surface

s gin
" ! #
Σt := x | F (x, y, z,t) = e−2t x2 + y2 + z2 e−t −C = 0 ,
2

t d le En

r
ba
where C = 0 is a constant, determine ϕ (t) considering that the particles on this

ge ro or
eS m
ci
surface are always the same.
f

ra
C d P cs
b
a
i
an an n

Solution
y ha

The function F defines the material surface Σt := {x | F (x, y, z,t) = 0}. The nec-
le
liv or ec

essary and sufficient condition for this surface to be a material surface is


M

.A

dF ∂F
= + v · ∇F = 0 ∀x ∈ Σt ∀t ,
∂t
m

dt
d
uu
e

where
!
X Th

∂F
er
tin

= −2e−2t x2 + y2 − 2tz2 e−t ,


2

∂t $ %
on

2 T
.O

∇F ≡ 2xe−2t , 2ye−2t , 2ze−t


not
, and
C

v · ∇F = 2x2 e−2t + 2y2 e−2t + 2z2 e−t ϕ (t) .


2
©

Then, the necessary and sufficient condition above is reduced to

2z2 (ϕ (t) − t) e−t = 0


2
∀x ∈ Σt ∀t .

Moreover, for x ∈ Σt , the term z2 can be isolated from the expression of


the function defining the material surface F (x, y, z,t) given in the statement,
!! 2
z2 = C − e−2t x2 + y2 et . Replacing this expression into the previous equa-
tion yields
!!
2 C − e−2t x2 + y2 (ϕ (t) − t) = 0 ∀x ∀t .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
34 C HAPTER 1. D ESCRIPTION OF M OTION

!!
Since C − e−2t x2 + y2 = 0 cannot be satisfied for ∀x and ∀t because C is a
constant, the only possibility left is

ϕ (t) = t .

Problem 1.5 – Given the velocity field of a perfect fluid



T
y

rs
not
v (x,t) ≡ ze ,t
, vz
1+t

ee
s gin
and the surface ϕ (x,t) = x − z (1 + t) e t + k = 0 (where k is a constant), which
is known to be a material surface, determine:

t d le En
a) The equation of the trajectory in canonical form and the equation of the

r
streamlines.

ba
ge ro or
eS m
ci
f
b) The equation of the streakline and the position of its initial and final points

ra
if the spill point is x∗ and the spill period is t ∈ [t1 ,t2 ].
C d P cs
b
a
i
an an n
y ha

le
Solution
liv or ec

a) To be able to calculate the trajectories and streamlines, the expression for


M

.A

the velocity field must be completed. To find vz , the information given about
surface ϕ is used. The necessary and sufficient condition for this surface to be a
m

material surface is
uu
e

∂ϕ
X Th


er
tin

= + v · ∇ϕ = 0 ∀x ∈ Σt ∀t ,
dt ∂t
on

.O

∂ϕ
∇ϕ ≡ [1 , 0 , −e t (1 + t)]T
C

not
where = −z (e t + e t (1 + t)) ,
∂t
©

and v · ∇ϕ = ze t − vz e t (1 + t) .

Then, the material derivative of ϕ is



= −ze t − ze t (1 + t) + ze t − vz e t (1 + t) = 0
dt
which results in vz = −z. Therefore, the spatial description of velocity field is

T
not y
v (x,t) ≡ ze ,t
, −z .
1+t

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 35

Now, this field must be integrated to obtain the equation of the trajectory since
dx/dt = v (x,t). Applying the equality for each component and particularizing
for the velocity field determined yields
dx dy y dz
= ze t , = and = −z .
dt dt 1+t dt
Note that the x-component depends on the z-coordinate. Then, the z-coordinate
must be determined first,
dz
= −z =⇒ z = C1 e−t .

rs
dt

ee
Replacing the expression found for z into the x-component and integrating the

s gin
expression results in

t d le En
dx
= C1 e−t e t = C1 =⇒ x = C1t +C2 .

r
dt

ba
ge ro or
eS m
ci
Finally, the y-component is
f

ra
C d P cs
b
a
dy y
= =⇒ y = C3 (1 + t) .
i
an an n

dt 1+t
y ha

To obtain the canonical form of the expression, x = X for t = 0 is imposed,


le
liv or ec


M

.A

⎨ x (0) = C2 = X

y (0) = C3 = Y
m



d

z (0) = C1 = Z
uu
e
X Th

er
tin

and, finally, the equation of the trajectory in canonical form is


on

.O

x = X + Zt
C

y = Y (1 + t) .
©

z = Ze−t

The equation of the streamlines is found by integrating the velocity field with
respect to λ , that is, dx (λ )/dλ = v (x (λ ) ,t). As in the case of the equation of
the trajectory, the z-component must be determined before the x-component,
dz
= −z =⇒ z = C1 e−λ .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
36 C HAPTER 1. D ESCRIPTION OF M OTION

Replacing into the x-component yields


dx
= C1 e (t−λ ) =⇒ x = −C1 e (t−λ ) +C2

and the remaining component results in
dy y λ
= =⇒ y = C3 e 1+t .
dλ 1+t
Then, the equation of the streamlines is

rs
ee
x = −C1 e (t−λ ) +C2

s gin
λ
y = C3 e 1+t .
z = C1 e−λ

t d le En

r
ba
ge ro or
eS m
ci
b) To obtain the equation of the streakline it is enough to take the equation of
f

ra
motion and impose x∗ = x (X, τ), where τ is a time belonging to the spill period.
C d P cs
b
a

i
an an n


⎨ x = X + Zτ

y ha

y∗ = Y (1 + τ)
le

liv or ec

⎩ ∗
z = Ze−τ
M

.A

And the inverse of this equation is


m


uu

X = x∗ − Zτ = x∗ − z∗ τeτ
e




X Th

y∗
er
tin

Y=

⎪ 1+τ

on

.O

Z = z ∗ eτ
C

Replacing these into the equation of motion results in the equation of the streak-
line,
x = x∗ − z∗ (τ − t) eτ
1+t
y = y∗ .
1+τ
z = z∗ e(τ−t)

Consider the physical concept of the streakline as the color line that would be
observed in the medium if a tracer fluid were injected at the spill point through-

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 37

out the spill period. Then, for each time t, the streakline can be visualized in
terms of the parameter τ, which gives the position in space of the colored parti-
cles. It is verified that, as expected, x = x∗ for t = τ, since it corresponds to the
time in which the streakline is crossing the spill point. Now, the streakline must
be delimited for each time t.

There are two distinct cases:


i) t1 < t < t2
The first colored point in the streakline is the one crossing the spill point at

rs
τ = t1 while the last one is the one crossing the spill point at τ = t.

ee

⎪ x = x∗ − z∗ (t1 − t) e t1 ⎧
⎪ ∗
⎪ ⎨x = x

s gin
⎨ 1 + t
Initial point: y = y∗ Final point: y = y∗
⎪ + ⎪

t d le En
⎪ 1 t ⎩


1
z = z∗
⎩ z = z∗ e (t1 −t)

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

ii) t ≥ t2
.O
C

The first colored point in the streakline is the same as in the previous case,
©

τ = t1 , but the last point is now τ = t2 . The streakline has now “moved away”
from the spill point.

⎪ ⎧


⎪ x = x∗ − z∗ (t1 − t) e t1 ⎪
⎪ x = x∗ − z∗ (t2 − t) e t2
⎨ 1+t ⎨ 1+t
Initial point: y = y∗ Final point: y = y∗

⎪ 1 + t1 ⎪
⎪ 1 + t2

⎪ ⎩
⎩ z = z∗ e (t1 −t) z = z∗ e (t2 −t)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
38 C HAPTER 1. D ESCRIPTION OF M OTION

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 39

E XERCISES

1.1 – Justify if the following statements are true or false.


a) Two streamlines corresponding to a same instant of time can never cross each
other unless the velocity field at the cross point is zero.
b) Two different trajectories can never cross each other.
c) Two streaklines corresponding to two spill points with the same spill period

rs
can cross each other at one or more points.

ee
s gin
1.2 – Given the following velocity field in material description
not  T

t d le En
v ≡ Ae At X1 , BtX1 , CX3 ,

r
ba
ge ro or
with A, B and C constants, obtain its spatial description and the conditions A, B

eS m
and C must fulfill for the motion to be feasible for 0 < t < ∞.

ci
f

ra
C d P cs
b
a
1.3 – Tracer fluid is injected at point (1, 1, 1) of the interior of a fluid from time
i
an an n

t = 1 to time t = 2. If the equation of the streamlines is


y ha

le
x = C1 eλt , y = C2 eλt , z = C3 e2λt
liv or ec
M

.A

determine the equation of the streakline, indicating its initial and final points for
t = 5.
m

d
uu
e

1.4 – The spatial description of the velocity field of a fluid is


X Th

er
tin

not  T
v ≡ ye−t , ze t , 0 .
on

.O

Tracer fluid is injected on plane y = 0 at time t = 1. Obtain the spatial equation


C

of the stain along time.


©

1.5 – A certain motion is defined by the velocity field


z
vx = 2ax ; vy = −by ; vz = − .
t +c
Determine:
a) The equation of the trajectory in canonical form and the equation of the
streamlines.
b) The possible values of a, b and c such that the motion has physical sense for
t ∈ [0, ∞).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
40 C HAPTER 1. D ESCRIPTION OF M OTION

c) The spatial description of the material surface that, at time t = 1, was a


sphere with center at (0, 0, 0) and radius R (consider a = b = c = 1).

rs
ee
s gin
t d le En

r
ba
1.6 – A certain motion is defined by the velocity field

ge ro or
eS m
ci
vx = ye−t ; vy = y ;f vz = 0 .

ra
C d P cs
b
a
i
Determine:
an an n
y ha

a) The equation of the trajectory in canonical form and the equation of the
le
streamlines.
liv or ec

b) The spatial description of the material surface that, at time t = 1, was a


M

.A

sphere with center at (0, 0, 0) and radius R.


m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
CH.2. DEFORMATION AND
STRAIN
Multimedia Course on Continuum Mechanics
Overview
 Introduction Lecture 1
 Deformation Gradient Tensor
 Material Deformation Gradient Tensor Lecture 2
 Inverse (Spatial) Deformation Gradient Tensor
Lecture 3

 Displacements Lecture 4
 Displacement Gradient Tensors
 Strain Tensors
 Green-Lagrange or Material Strain Tensor Lecture 5
 Euler-Almansi or Spatial Strain Tensor
 Variation of Distances
 Stretch Lecture 6
 Unit elongation
 Variation of Angles Lecture 7

2
Overview (cont’d)
 Physical interpretation of the Strain Tensors Lecture 8
 Material Strain Tensor, E
 Spatial Strain Tensor, e Lecture 9
 Polar Decomposition Lecture 10
 Volume Variation Lecture 11
 Area Variation Lecture 12
 Volumetric Strain Lecture 13
 Infinitesimal Strain
 Infinitesimal Strain Theory
 Strain Tensors
 Stretch and Unit Elongation Lecture 14
 Physical Interpretation of Infinitesimal Strains
 Engineering Strains
 Variation of Angles

3
Overview (cont’d)
 Infinitesimal Strain (cont’d)
 Polar Decomposition Lecture 15
 Volumetric Strain
 Strain Rate
 Spatial Velocity Gradient Tensor Lecture 16
 Strain Rate Tensor and Rotation Rate Tensor or Spin Tensor
 Physical Interpretation of the Tensors
 Material Derivatives Lecture 17
 Other Coordinate Systems
 Cylindrical Coordinates Lecture 18
 Spherical Coordinates

4
2.1 Introduction
Ch.2. Deformation and Strain

5
Deformation
 Deformation: transformation of a body from
a reference configuration to a current configuration.

 Focus on the relative movement of a given particle w.r.t. the particles


in its neighbourhood (at differential level).

 It includes changes of size and shape.

6
2.2 Deformation Gradient Tensors
Ch.2. Deformation and Strain

7
Continuous Medium in Movement
Ω0: non-deformed (or reference) Ω or Ωt: deformed (or present)
configuration, at reference time t0. configuration, at present time t.
X : Position vector of a particle at x : Position vector of the same particle
reference time. at present time.

ϕ ( X,t )

t0 Q t
Reference or Ωt
dX Q’
non-deformed dx

Ω0 P P’
X x Present or
deformed

8
Fundamental Equation of Deformation
 The Equations of Motion:
not
xi i ( X1, X 2 , X 3 , t )
ϕ= xi ( X 1 , X 2 , X 3 , t ) i ∈ {1, 2,3}
not
= ( X, t ) x ( X, t )
x ϕ=

 Differentiating w.r.t. X :
 ∂xi ( X, t ) not
 dxi = dX j Fij ( X, t ) dX j i, j ∈ {1, 2,3}
 ∂X j

 ∂x ( X, t ) not Fundamental equation
 d=
x X F ( X, t ) ⋅ dX
⋅ d= of deformation
∂X

(material) deformation
gradient tensor

9
Material Deformation Gradient Tensor

 The (material) deformation gradient tensor: REMARK  ∂ 


 
The material Nabla operator  ∂X 1 
 not  ∂ 
F (=
X, t ) x( X, t ) ⊗ ∇ is defined as: ∇ ≡ ∂ eˆ i ∇  = 
∂X i  ∂X 2 
 not ∂x  ∂ 
=  Fij
i
i, j ∈ {1, 2,3}  
 ∂X 3 
 ∂X j
 ∂x1 ∂x1 ∂x1 
 
 x1   ∂X 1 ∂X 2 ∂X 3 
 ∂ ∂ ∂   ∂x2 ∂x2 ∂x2 
[F ] = x ⊗ ∇  = x2  

 = 
∂X ∂X 2 ∂X 3   ∂X 1 ∂X 2 ∂X 3 
 x3   1
 ∂x3 ∂x3 ∂x3 
=  x  =∇ T  
 ∂X 1 ∂X 2 ∂X 3 

 F(X,t):
 is a primary measure of deformation
 characterizes the variation of relative placements in the neighbourhood of a
material point (particle).
=dx F ( X, t ) ⋅ dX
10
Inverse (spatial) Deformation Gradient
Tensor
 The inverse Equations of Motion:
not
( x1 , x2 , x3 , t ) X i ( x1 , x2 , x3 , t )
X i ϕ= i
−1
i ∈ {1, 2,3}
not
= ( x, t ) X ( x, t )
X ϕ= −1

 Differentiating w.r.t. x :
∂X i ( X, t ) not
=dX i = dx j Fij−1 ( x, t ) dx j i, j ∈ {1, 2,3}
∂x j
∂X ( x, t ) not
=
dX =⋅ dx F −1 ( x, t ) ⋅ dx
∂x

Inverse (spatial) deformation


gradient tensor

11
Inverse (spatial) Deformation Gradient
Tensor
 The spatial (or inverse) deformation gradient tensor:
= dX F −1 ( x, t ) ⋅ dx

∂
 F ( x, t ) ≡ X(x, t ) ⊗ ∇
−1 REMARK x 
 The spatial Nabla  1
 −1 ∂X i ∂
=
 Fij i, j ∈ {1, 2,3} operator is defined as: [∇] = 
 ∂x j ∂  x2 
∇≡ eˆ i ∂
 ∂X 1 ∂X 1 ∂X 1  ∂xi
   
= ∇T  ∂x1 ∂x2 ∂x3   x3 
 X1 
  ∂ ∂ ∂   ∂X 2 ∂X 2 ∂X 2 
F  = [ X ⊗ ∇ ] =  X 2   ∂x
−1
= 
∂x2 ∂x3   ∂x1 ∂x2 ∂x3 
 X 3   1
 ∂X 3 ∂X 3 ∂X 3 
 F-1(x,t): =  X   
 ∂x1 ∂x2 ∂x3 
 is a primary measure of deformation
 characterizes the variation of relative placements in the neighbourhood of a
spatial point.
 It is not the spatial description of the material deformation gradient tensor

12
Properties of the Deformation
Gradients
 The spatial deformation gradient tensor is the inverse of the material
deformation gradient tensor:

∂xi ∂X k ∂xi
= = δ ij F ⋅ F −1 = F −1 ⋅ F = 1
∂X k ∂x j ∂x j

 If F is not dependent on the space coordinates, F( X, t ) ≡ F ( t ) the


deformation is said to be homogeneous.
 Every part of the solid body deforms as the whole does.
 The associated motion is called affine.

∂x
 If there is no motion, =
x X and =
F −1
= F= 1 .
∂X

13
Example
Compute the deformation gradient and inverse deformation gradient tensors
for a motion equation with Cartesian components given by,
 X + Y 2t 
[ x] Y (1 + t ) 
=
 Zet 
 

1.
Using the results obtained, check that F ⋅ F −1 =

14
 X + Y 2t 
 
[ ]  ( )
Example - Solution =x Y 1 +
 Zet 

t

The Cartesian components of the deformation gradient tensor are,


 X + Y 2t  1 2Yt 0 
  ∂ ∂ ∂   
F ( X, t )  =x ⊗ ∇ ≡ [ x ] ∇  =Y (1 + t )  
T
, , = 0 1 + t 0
 ∂X ∂Y ∂Z   
 Ze  t
0 0 e  t
 

The Cartesian components of the inverse motion equation will be given by,

 y 2t 
 X= x −   2 yt 
( + )
2
 1 t   1 0
  (1 + t )
 y     
[ X=] ϕ −1 ( x, t =
)  =
   f (=
( ) x, t )  0 1 + t
Y  F X ( x , =
t ), t 0
 1+ t 
  0 et 
 Z = ze − t   f ( x ,t )  
0

 
   

15
Example - Solution
The Cartesian components of the inverse deformation gradient tensor are,
 2 yt 
 1 − 0
( + )
2
 1 t 
 1 
F −1 ( x, t )  = 0 0
 1+ t 
 
 
0 0 e − t 

1:
And it is verified that F ⋅ F −1 =
 2 yt   2 yt 2 yt 
  1 − 0 1 − + 0 
0  
2 yt
(1+ t) (1 + t ) (1 + t )
2 2 2
 1
(1 + t )     1 0 0
   1   1+ t  0 1 0= 
F ⋅ F −=
1
0 1 + t 0  ⋅ 0 0=  0 0 =    [1]
0 1+ t 1+ t
0 et      0 0 1 
     
  0  0 
 0 e−t   0 et e − t 

16
2.3 Displacements
Ch.2. Deformation and Strain

17
Displacements
 Displacement: relative position of a particle, in its current (deformed)
configuration at time t, with respect to its position in the initial
(undeformed) configuration.
 Displacement field: displacement of all the particles in the continuous
medium.
 Material description (Lagrangian form):
( X, t ) x ( X, t ) − X
U= t0 t

U i ( X, t ) = xi ( X, t ) − X i i ∈ {1, 2,3} Ω
P U=u P’
 Spatial description (Eulerian form): Ω0 X
x
u ( x, t )= x − X ( x, t )
ui ( x, t ) =
xi − X i ( x, t ) i ∈ {1, 2,3}

18
Displacement Gradient Tensor
U= ( X, t ) x ( X, t ) − X  Taking partial derivatives of U w.r.t. X :
 ∂U i ( X, t ) ∂xi ( X, t ) ∂X i
U i ( X , t ) = xi ( X , t ) − X i i ∈ {1, 2,3}
def
= − = Fij − δ ij = J ij
∂X j ∂X j ∂X

  j
 ∂U i δij
 J = = Fij − δ ij i, j ∈ {1, 2,3} Fij
∂X j
ij

 Material Displacement
 def
Gradient Tensor
 (
J X , t ) = U ( X, t ) ⊗ ∇ = F − 1
 Taking partial derivatives of u w.r.t. x :
u ( x, t )= x − X ( x, t ) ∂ui (x, t ) ∂xi ∂X i (x, t ) def
 = − = δ ij − Fij − δ ij = jij
−1

ui ( x, t ) =
xi − X i ( x, t ) i ∈ {1, 2,3} ∂x j ∂x j
 
∂x j
 
δij Fij−1
 ∂ui
 ij ∂x =
j = δ ij − Fij−1 i, j ∈ {1, 2,3} Spatial Displacement Gradient
 j
Tensor
 def
 j ( x, t ) = u ( x, t ) ⊗ ∇ =1 − F
−1

REMARK If motion is a pure shifting: ∂x


x( X, t ) =X + U(t ) ⇒ F= = 1=F −1 and j =J=0
. ∂X
19
2.4 Strain Tensors
Ch.2. Deformation and Strain

20
Strain Tensors
 F characterizes changes of relative placements during motion but is not
a suitable measure of deformation for engineering purposes:
 It is not null when no changes of distances and angles take place, e.g.,
in rigid-body motions.

 Strain is a normalized measure of deformation which characterizes the


changes of distances and angles between particles.
 It reduces to zero when there is no change of distances and angles
between particles.

21
Strain Tensors
Consider
F ( X,t )

t0 Q t d=
x F ⋅ dX
Ω dxi = Fij dX j
dS dX dx Q’
ds
d=
X F -1 ⋅ dx
Ω0 P P’
X x dX i = Fij−1dx j

 where dS is the length of segment dX : =dS dX ⋅ dX


 and ds is the length of segment dx : =
ds dx ⋅ dx

22
Strain Tensors
d=
X F -1 ⋅ dx d=
x F ⋅ dX
dX i = Fij−1dx j dxi = Fij dX j

=
dS dX ⋅ dX =
ds dx ⋅ dx

 One can write:


( ) = ⋅ = [ ] ⋅ [ ] = [ ⋅ ] ⋅ [ F ⋅ dX ] = dX ⋅ FT ⋅ F ⋅ dX
2 T T
 ds dx dx dx dx F dX

( ds ) dx
= = = =
2 T
k dx k F ki dX F
i kj dX j dX F F
i ki kj dX j dX i ik Fkj dX j
F
REMARK
 not
( dS ) = dX ⋅ dX =[ dX ] ⋅ [ dX ] = F ⋅ dx  ⋅ F ⋅ dx  = dx ⋅ F ⋅ F ⋅ dx
2 T −1 T −1 −T −1 The convention
not

T
(•) −1  =
(•) −T
( dS )2 dX
= = k dX k Fki−1 dxi=
Fkj−1 dx j dxi Fki−=
1 −1
Fkj dx j dxi Fik−T Fkj−1dx j is used.

23
Green-Lagrange Strain Tensor
( ds ) = dX ⋅ FT ⋅ F ⋅ dX ( dS =
) dX ⋅ dX
2 2

 Subtracting:
( ds ) − ( dS ) = dX ⋅ FT ⋅ F ⋅ dX − dX ⋅ dX = dX ⋅ FT ⋅ F ⋅ dX − dX ⋅ 1 ⋅ dX = dX ⋅ ( FT ⋅ F − 1 ) ⋅ dX = 2 dX ⋅ E ⋅ dX
2 2


def
= 2E

 The Green-Lagrange or Material Strain Tensor is defined:



) ( F ⋅ F − 1)
1 T
 (
E X,=t
2

 E ( X, t ) = 1 ( F F − δ ) i, j ∈ {1, 2,3}
 ij 2
ki kj ij

( )
 E is symmetrical:
(2
1 T
) 2 F ⋅ ( F ) − 1= 2 ( F ⋅ F − 1=) E
1 T T T T 1 T
T
E=
T
F ⋅ F − 1=

=Eij E ji i, j ∈ {1,2,3}

24
Euler-Almansi Strain Tensor
( ds )= dx ⋅ dx ( dS ) =dx ⋅ F −T ⋅ F −1 ⋅ dx
2 2

 Subtracting:
( ds ) − ( dS ) = dx ⋅ dx − dx ⋅ F −T ⋅ F −1 ⋅ dx = dx ⋅ 1 ⋅ dx − dx ⋅ F −T ⋅ F −1 ⋅ dx =
2 2
def
= 2e
= dx ⋅ (1 − F ⋅ F ) ⋅ dx = 2 dx ⋅ e ⋅ dx
−T −1

  
def
= 2e
 The Euler-Almansi or Spatial Strain Tensor is defined:

 e ( x, t ) =
1
2
( 1 − F −T
⋅ F −1
)

 e ( x, t ) =
 ij
1
2
( δ ij − Fki−1 Fkj −1 ) i, j ∈ {1, 2,3}

 e is symmetrical: e =
T 1
2
( −T −1 T 1 T
2
−1 T
( −T T 1
2
)
1 − F ⋅ F ) = 1 − ( F ) ⋅ ( F ) = (1 − F −T ⋅ F −1 ) = e

= eij e ji i, j ∈ {1, 2,3}


25
Particularities of the Strain Tensors
 The Green-Lagrange and the Euler-Almansi Strain Tensors are different
tensors.
 They are not the material and spatial descriptions of a same strain tensor.
 They are affected by different vectors (dX and dx) when measuring distances:

( ds ) − ( dS )= 2 dX ⋅ E ⋅ dX
= 2 dx ⋅ e ⋅ d x
2 2

 The Green-Lagrange Strain Tensor is inherently obtained in material


description, E = E ( X,t ) .
 =
By substitution of the inverse Equations of ( X ( x, t ) , t ) E ( x, t ) .
Motion, E E=

 The Euler-Almansi Strain Tensor is inherently obtained in spatial description,


e = e ( X,t ) .
 =
By substitution of the Equations of ( x ( X, t ) , t ) e ( X, t ) .
Motion, e e=

26
Strain Tensors in terms of Displacements
 Substituting F −1= 1 - j and F= J + 1 into
E=
2
( F ⋅ F − 1) and e= 1 (1 − F −T ⋅ F −1 ) :
1 T
2

 1 1
 E= (1 + J T ) ⋅ (1 + J ) − 1=  J + J T + J T ⋅ J 
 2 2

 Eij= 1  ∂U i ∂U j ∂U k ∂U k 
 + +  i, j ∈ {1, 2,3}
 2  ∂X j ∂X i ∂X i ∂X j 

 1
1 − (1 − jT ) ⋅ (1 − j) =  j + jT − jT ⋅ j
1
 e=  2 
 2
  ∂u 
 eij= 1  ∂ui + j − ∂uk ∂uk  i, j ∈ {1, 2,3}
 2  ∂x j ∂xi ∂xi ∂x j 

27
Example
For the movement in the previous example, obtain the strain tensors in the
material and spatial description.
 X + Y 2t 
[ x] Y (1 + t ) 
=
 Zet 
 

29
Example - Solution
The deformation gradient tensor and its inverse tensor have already been
obtained:  2 yt 
 1 − 0 
( + )
2
 1 t 
1 2Yt 0 
0 1 + t 0   1 
F=
  F −1 =
0 0
 1+ t 
0 0 e t
 
 
0 0 e − t 

The material strain tensor :


 1 0 0  1 2Yt 0   1 − 1 2Yt 0 
1  1 
=
E
2
(
1 T
F ⋅F −1 =) 2  
2Yt 1 + t 0  ⋅  0 1 + t 0  − 1=
    2  2Yt ( 2Yt ) + (1 + t ) − 1
2 2
0 = 
  0 e e − 1
  0 0 et  0 0 et  t t
 0
 
 0 2Yt 0 
1 
( ) + ( + ) −
2 2
 2Yt 2Yt 1 t 1 0 
2
0 0 e − 1
2t
 

30
Example - Solution
The spatial strain tensor :
 2 yt 
  2 yt   1−1 − 0 
1 − (1 + t )
2
  0 
 0  (1 + t )  
2
 
1 0
   
   1 2 
2
1    1  2 yt =
e=
1
(1 − F−T ⋅ F −1 )=  2 yt
1 − −
1
0  ⋅ 0

1
0  = −
 (1 + t ) 2
1− −
2 yt
 + 
  (1 + t ) 2   1 + t  
0

(1+ t) 1+ t 1+ t
2
2 2  2
        
  e − t  0 0 e−t    −t −t 

0 0
   0 0 1− e e
 
    
 2 yt 
 0 − 0 
(1 + t )
2
 
 2
 2 yt   1 2 
1  2 yt
= − 1−   − 0 
2  (1 + t ) 2  (1 + t ) 2   1 + t  
   
 0 0 1 − e −2 t 
 
 

31
 X + Y 2t 
[ x] Y (1 + t ) 
Example - Solution =
 Zet 
 

In conclusion, the material strain tensor is: spatial description


 2 yt 
material description  0 (1 + t )
0 
 0 2Yt 0   
1    2 yt 
2 
E ( X, t ) ( 2Yt ) + (1 + t ) − 1 0  1  2 yt
E ( x, t )  + (1 + t ) − 1 0 
2 2
=
2
 2Yt 
2 2  (1 + t )  (1 + t )  
0 0 e 2t − 1 Y=
y
 
  (1 + t )  0 0 e 2t − 1
 
And the spatial strain tensor is:  

 2 yt   
 0 − 0  −
2Yt
(1 + t )
2
 0 (1 + t )
0 
   
 2   
1  2 yt  2 yt   1 2   2Yt   1 
2 2
e ( x, t ) =
1  2Yt
 − 1−   −  0  e ( X, t ) = − 1 −   −   0 
2 (1 + t )  
 (1 + t )   1 + t  2  (1 + t )  (1 + t )   1 + t  
2 2
  y Y (1 + t )
=  
 0 0 1− e −2 t 
 0 0 1 − e −2t 
   
   
spatial description material description
Observe that E ( x, t ) ≠ e ( x, t ) and E ( X, t ) ≠ e ( X, t ).
32
2.5 Variations of Distances
Ch.2. Deformation and Strain

33
Stretch
 The stretch ratio or stretch is defined as:
def P´Q´ ds
stretch = λ T= λ=
t = (0 < λ < ∞)
PQ dS

t0 Q t t

dS dX Ω Q’ REMARK
dx
ds The sub-indexes (●)T and
(●)t are often dropped. But
Ω0 P P’
X x one must bear in mind that
stretch and unit elongation
always have a particular
direction associated to them.

34
Unit Elongation
 The extension or unit elongation is defined as:
def ∆ PQ ds − dS
unit elongation = ε=
T ε=
t =
PQ dS

t0 Q t t

dS dX Ω Q’ REMARK
dx
ds The sub-indexes (●)T and
(●)t are often dropped. But
Ω0 P P’
X x one must bear in mind that
stretch and unit elongation
always have a particular
direction associated to them.

35
Relation between Stretch and Unit
Elongation
 The stretch and unit elongation for a same point and direction are
related through:
ds − dS ds
ε= = − 1= λ − 1 ( −1 < ε < ∞ )
dS dS

 = (ε 0 ) =
If λ 1= ds dS : P and Q may have moved in time but have kept
the distance between them constant.

 If λ > 1 ( ε > 0 ) ds > dS : the distance between them P and Q has


increased with the deformation of the medium.

 If λ < 1 ( ε < 0 ) ds < dS : the distance between them P and Q has


decreased with the deformation of the medium.

36
Stretch and Unit Elongation in terms of
the Strain Tensors
 Considering:
( ds ) − ( dS )= 2 dX ⋅ E ⋅ dX
2 2
dX = T dS
( ds ) − ( dS )= 2 dx ⋅ e ⋅ dx dx = t ds
2 2

 Then:
× 1
( dS ) 2 = λ2
2
 ds 
( ds ) − ( dS ) = 2 ( dS ) T ⋅ E ⋅ T   − 1= 2 T ⋅ E ⋅ T
2 2 2

 dS  REMARK
2
 dS  E(X,t) and e(x,t)
( ds ) − ( dS ) = 2 ( ds ) t ⋅ e ⋅ t 1−   = 2 t ⋅e ⋅ t
2 2 2

 ds  contain information
× 1
( λ)
2
( ds )2 = 1 regarding the stretch
and unit elongation for
1
λ= any direction in the
λ = 1+ 2 T ⋅ E ⋅ T 1− 2 t ⋅e ⋅ t differential neighbour-
ε = λ −1 = 1 + 2 T ⋅ E ⋅ T −1 1 hood of a point.
ε = λ −1 = −1
1− 2 t ⋅e ⋅ t
37
2.6 Variation of Angles
Ch.2. Deformation and Strain

38
Variation of Angles
dx( ) = t ( ) ds ( )
1 1 1
(1) (1) (1)
dX = T dS
dx( ) = t ( ) ds (
2 2 2)
( 2) ( 2) ( 2) T(1)
dX =T dS t
t0
Q t(1)
t(2)
T(2)
Θ Ω R’ θ
dS(1) Q’
R ds(2) ds(1)
dS(2)
Ω0 P P’
X x

 The scalar product of the vectors dx(1) and dx(2) :

dx(1) ⋅ dx( 2) = dx(1) ⋅ dx( 2) cos θ = ds (1) ds ( 2) cos θ

39
Variation of Angles
d x( 2) =
x(1) ⋅ d ds (1) ds ( 2) cos θ
 dx(1) 
T
 dx ( 2 ) 
   

(1)
dx = F ⋅ dX (1) T(1) ⋅ (1 + 2E ) ⋅ T( 2)
cos θ =
 ( 2) ( 2) 1 + 2 T(1) ⋅ E ⋅ T(1) 1 + 2 T( 2 ) ⋅ E ⋅ T( 2 )
dx = F ⋅ dX
=  F ⋅ dX (1)  ⋅  F ⋅ dX (2)  = dX (1) ⋅ ( FT ⋅ F ) ⋅ dX (2)
T
(1) (2)
dx ⋅ dx
 
2E+1
dX(1) = T(1) dS (1) 

dX( ) = T( ) dS ( ) 
2 2 2

 1 1 
dx (1)=
⋅ dx (2)  T ⋅ ( 2E + 1 ) ⋅ T = ds (1)ds (2)  (1) (2) T(1) ⋅ ( 2E + 1) ⋅=
(1) (1) (2) ( 2)
dS 
dS T(2)  ds (1)ds (2) cos θ
ds(1) ds(2) λ λ 
(1) (2)
λ λ
λ = 1+ 2 T ⋅ E ⋅ T

 T(1) ⋅ (1 + 2E ) ⋅ T( 2) 
 
(1) (1)
 1+ 2 T ⋅ E ⋅ T 1 + 2 T( 2 ) ⋅ E ⋅ T( 2 ) 
40
Variation of Angles

dX(1) = T(1) dS (1) dx(1) = t (1) ds (1)


 ( 2)  ( 2) ( 2) ( 2)
( 2) ( 2) T(1) dx = t ds
dX = T dS t
t 0
Q t(1)
t(2)
T(2)
Θ Ω R’ θ
dS(1) Q’
R ds(2) ds(1)
dS(2)
Ω0 P P’
X x

 The scalar product of the vectors dX(1) and dX(2) :


X( 2) dX(1) ⋅ dX( 2) cos
dX(1) ⋅ d= = Θ dS (1) dS ( 2) cos Θ

41
Variation of Angles
X(1) =
d X( 2) dS (1) dS ( 2) cos Θ
⋅ d
 dX(1) 
T
 dX( 2) 
   
t (1) ⋅ (1 − 2e ) ⋅ t ( 2)
dX=(1)
F −1 ⋅ dx(1) cos Θ =
 ( 2) −1
dX= F ⋅ dx
( 2) 1 − 2 t (1) ⋅ e ⋅ t (1) 1 − 2 t ( 2) ⋅ e ⋅ t ( 2)
dX (1) ⋅ dX (2) =  F −1 ⋅ dx (1)  ⋅  F −1 ⋅ dx (2)  = dx (1) ⋅ ( F −T ⋅ F −1 ) ⋅ dx (2)
T


1−2e
dx(1) = t (1) ds (1)
 ( 2) ( 2) ( 2)
dx = t ds

dX (1) ⋅ dX (2)= (1) (1) (2)


 t ⋅ (1 - 2e ) ⋅ t ds
ds
(1) (1)
( 2) (1) (2) (1) ( 2 ) (1)
(2) (2)
(2)
((1) (2)
 = dS dS λ λ t ⋅ (1 - 2e ) ⋅ t = dS dS cos Θ )
λ dS λ dS

REMARK λ=
1
E(X,t) and e(x,t) contain information 1− 2 t ⋅e ⋅ t
regarding the variation in angles  t (1) ⋅ (1 − 2e ) ⋅ t ( 2) 
between segments in the differential  
(1) (1)
neighbourhood of a point.  1− 2 t ⋅e ⋅ t 1 − 2 t ( 2) ⋅ e ⋅ t ( 2) 
42
Example
Let us consider the motion of a continuum body such that the spatial description
of the Cartesian components of the spatial Almansi strain tensor is given by,
 0 0 −tetz 
 
[ e ( x, t ) ] =  0 0 0 
 tz 
 −te 0 t ( 2etz − et ) 

Compute at time t=0 (the reference time), the length of the curve that at time
t=2 is a straight line going from point a (0,0,0) to point b (1,1,1).
The length of the curve at time t=0 can be expressed as,
1
∫ λ ( x, t ) ds
B b
=L ∫=
A 
dS
a
= ds
λ
43
Example - Solution
The inverse of the stretch, at the points belonging to the straight line going
from a(0,0,0) to b(1,1,1) along the unit vector in the direction of the straight
line, is given by,
1
λ ( x, t ) = → λ −1 ( x, t ) = 1 − 2t ⋅ e ( x, t ) ⋅ t
1 − 2t ⋅ e ( x, t ) ⋅ t
Where the unit vector is given by,
1
[t ] = [1 1 1]
T

3
Substituting the unit vector and spatial Almansi strain tensor into the expression
of the inverse of the stretching yields,
2 t
λ −1
( x, t=) 1 + te
3

44
Example - Solution
The inverse of the stretch, which is uniform and therefore does not depends on
the spatial coordinates, at time t=2 reads,
4
λ −1 ( x, 2=
) 1 + e2
3

Substituting the inverse of the stretch into the integral expression provides the
length at time t=0,
4 2 4 2 (1,1,1)
( )
b b
∫Γ ∫a λ ∫a 3 ∫
−1
L= dS = x , 2 ds = 1 + e ds =1 + e ds =+
3 4e 2

3  
(0,0,0)

= 3

45
2.7 Physical Interpretation of E and e
Ch.2. Deformation and Strain

46
Physical Interpretation of E
 Consider the components of the material strain tensor, E:
 E XX E XY E XZ   E11 E12 E13 
= 
E = E XY EYY EYZ  E E22 E23 
 12
 E XZ EYZ EZZ   E13 E23 E33 

 For a segment parallel to the X-axis, the stretch is:


λ = 1+ 2 T ⋅ E ⋅ T
 E11 E12 E13  1 
reference T(1) ⋅ E
= ⋅ T(1) [1 0 0] ⋅  E12 E22 E23  ⋅=0 E
      11
configuration
 E13 E23 E33  0
T
T(1) 
  
T(1) 
 

Stretching of
1  dS  λ= 1 + 2E11 the material in
    1
T (1)
≡ 0  dX ≡ dS T (1)
=
0
0  0
the X-direction
   

47
Physical Interpretation of E
 Similarly, the stretching of the material in the Y-direction and the Z-
direction:
λ1 = 1 + 2 E11 ε X = λ X − 1= 1 + 2 E XX − 1
λ2 = 1 + 2 E22 ε Y = λY − 1= 1 + 2 EYY − 1
λ3 = 1 + 2 E33 ε Z = λZ − 1= 1 + 2 EZZ − 1

 The longitudinal strains contain information on the stretch and unit


elongation of the segments initially oriented in the X, Y and Z-directions
(in the material configuration).
 E XX E XY E XZ 
If E XX 0=
= εX 0 No elongation in the X-direction
E =  E XY EYY EYZ 
 E XZ EYZ EZZ  If EYY 0=
= εY 0 No elongation in the Y-direction
=
If EZZ 0=εZ 0 No elongation in the Z-direction

48
Physical Interpretation of E
 Consider the angle between a segment parallel to the X-axis and a
segment parallel to the Y-axis, the angle is:
T ( ) ⋅ ( 1 + 2E ) ⋅ T ( )
1 2

cos θ =
1 + 2 T(1) ⋅ E ⋅ T(1) 1 + 2 T( 2 ) ⋅ E ⋅ T( 2 )

reference 1  0  T(1) ⋅ T( 2) =
0
    T ⋅ E ⋅ T(1) =
(1)
configuration T(1) = 0  T( 2) = 1  E11
0 
 
0 
  T(1) ⋅ E ⋅ T( 2) =E12
( 2) ( 2)
T ⋅E ⋅T = E22
2 E12
cos θ =
1 + 2 E11 1 + 2 E 22

deformed
configuration

2 E XY π 2 E XY
θ ≡ θ xy =arccos = − arcsin
1 + 2 EXX 1 + 2 EYY 2 1 + 2 EXX 1 + 2 EYY

49
Physical Interpretation of E
π 2 E XY
θ ≡ θ xy = − arcsin
2 1 + 2 EXX 1 + 2 EYY

 The increment of the final angle w.r.t. its initial value:

2 E XY
θ xy − Θ
∆Θ XY =

=− arcsin
1 + 2 EXX 1 + 2 EYY
XY
π
2
reference
configuration

deformed
configuration

50
Physical Interpretation of E
 Similarly, the increment of the final angle w.r.t. its initial value for
couples of segments oriented in the direction of the coordinate axes:
2 E XY
∆Θ XY = − arcsin
1 + 2 EXX 1 + 2 EYY
2 E XZ
∆Θ XZ = − arcsin
1 + 2 EXX 1 + 2 EZZ
2 EYZ
∆ΘYZ = − arcsin
1 + 2 EYY 1 + 2 EZZ

 The angular strains contain information on the variation of the angles


between segments initially oriented in the X, Y and Z-directions (in the
material configuration). If E = 0 No angle variation between the
XY
 E XX E XY E XZ  X- and Y-directions
E =  E XY EYY EYZ  If E XZ = 0 No angle variation between the
 E XZ EYZ EZZ  X- and Z-directions
If EYZ = 0 No angle variation between the
Y- and Z-directions
51
Physical Interpretation of E
 In short,
deformed
configuration

reference λ1 dX= 1 + 2 E XX dX
configuration
λ2 dY= 1 + 2 EYY dY
λ3 dZ= 1 + 2 EZZ dZ

2 E XY
∆Θ XY = − arcsin
1 + 2 EXX 1 + 2 EYY
2 E XZ
∆Θ XZ = − arcsin
1 + 2 EXX 1 + 2 EZZ
2 EYZ
∆ΘYZ = − arcsin
1 + 2 EYY 1 + 2 EZZ

52
Physical Interpretation of e
 Consider the components of the spatial strain tensor, e:
 exx exy exz   e11 e12 e13 
   
e ≡ exy eyy eyz  =  e12 e22 e23 
 exz eyz ezz   e13 e23 e33 

 For a segment parallel to the x-axis, the stretch is:


1
λ=
1− 2 t ⋅e ⋅ t
deformed  e11 e12 e13  1 
configuration t ⋅ e ⋅ t [1 0 0] ⋅ e12
= e22 e23 =
⋅ 0  e11
 e13 e23 e33  0 

1  ds  1 Stretching of
    λ1 = the material in
t (1) ≡ 0  dx ≡  0  1 − 2e11
0  0 the x-direction
   
53
Physical Interpretation of e
 Similarly, the stretching of the material in the y-direction and the z-
direction: 1 1
λ1 = ⇒ ε x = λx − 1 = −1
1 − 2e11 1 − 2exx
1 1
λ2 = ⇒ ε y = λy − 1= −1
1 − 2e22 1 − 2eyy
1 1
λ3 = ⇒ ε z = λz − 1 = −1
1 − 2e33 1 − 2ezz

 The longitudinal strains contain information on the stretch and unit


elongation of the segments oriented in the x, y and z-directions (in the
deformed or actual configuration).
 exx exy exz 
 
e ≡ exy eyy eyz 
 exz eyz ezz 

54
Physical Interpretation of e
 Consider the angle between a segment parallel to the x-axis and a
segment parallel to the y-axis, the angle is:
t (1) ⋅ (1 − 2e ) ⋅ t ( 2)
cos Θ =
reference deformed 1 − 2 t (1) ⋅ e ⋅ t (1) 1 − 2 t ( 2) ⋅ e ⋅ t ( 2)
configuration configuration
1  0  t (1) ⋅ t ( 2) = 0
(1) (1)
    t ⋅ e ⋅t = e11
t (1) = 0  t (2) = 1 
0  0  t (1) ⋅ e ⋅ t ( 2) =e12
    ( 2)
t ⋅e ⋅t = ( 2)
e22
−2 e12
cos Θ =
1 − 2 e11 1 − 2 e 22

π 2exy
Θ ≡ Θ XY= + arcsin
2 1 − 2 exx 1 − 2 eyy

55
Physical Interpretation of e
π 2exy
Θ ≡ Θ XY= + arcsin
2 1 − 2 exx 1 − 2 eyy

 The increment of the angle in the reference configuration w.r.t. its value
in the deformed one:
2exy
∆θ xy = θ xy − Θ XY = − arcsin
 1 − 2 exx 1 − 2 eyy
π
2

reference deformed
configuration configuration

56
Physical Interpretation of e
 Similarly, the increment of the angle in the reference configuration w.r.t.
its value in the deformed one for couples of segments oriented in the
direction of the coordinate axes:
π 2exy
∆θ xy = − Θ XY = − arcsin
2 1 − 2 exx 1 − 2 eyy
π 2exz
∆θ xz = − Θ XZ = − arcsin
2 1 − 2 exx 1 − 2 ezz
π 2eyz
∆θ yz = − ΘYZ = − arcsin
2 1 − 2 eyy 1 − 2 ezz

 The angular strains contain information on the variation of the angles


between segments oriented in the x, y and z-directions (in the deformed
or actual configuration).  exx exy exz 
 
e ≡ exy eyy eyz 
 exz eyz ezz 

57
Physical Interpretation of e
 In short, reference
configuration
deformed
configuration
dx
= 1 − 2exx dx
λ1
dy
= 1 − 2eyy dy
λ2
dz
= 1 − 2ezz dz
λ3

π 2exy
∆θ xy = − Θ XY = − arcsin
2 1 − 2 exx 1 − 2 eyy
π 2exz
∆θ xz = − Θ XZ = − arcsin
2 1 − 2 exx 1 − 2 ezz
π 2eyz
∆θ yz = − ΘYZ = − arcsin
2 1 − 2 eyy 1 − 2 ezz

58
2.8 Polar Decomposition
Ch.2. Deformation and Strain

59
Polar Decomposition
 Polar Decomposition Theorem:
 “For any non-singular 2nd order tensor F there exist two unique
positive-definite symmetrical 2nd order tensors U and V, and a unique
orthogonal 2nd order tensor Q such that: ”
not
 left polar
=
U F ⋅F
T
 decomposition
not 
V = F ⋅ FT  F = Q ⋅ U = V ⋅Q
Q= F ⋅ U −1 =V −1 ⋅ F  right polar
 decomposition
 The decomposition is unique. REMARK
 Q: Rotation tensor An orthogonal 2nd
order tensor verifies:
 U: Right or material stretch tensor
QT ⋅ Q = Q ⋅ QT = 1
 V: Left or spatial stretch tensor

60
Properties of an orthogonal tensor
 An orthogonal tensor Q when multiplied (dot product) times
a vector rotates it (without changing its length): y= Q ⋅ x
 y has the same norm as x:
y = y ⋅ y = [ y ] [ y ] = [Q ⋅ x ] ⋅ [Q ⋅ x ] = x ⋅ Q
⋅Q⋅x = x
2 T T T 2

 when Q is applied on two vectors x(1) and x(2), with the same origin,
the original angle they form is maintained:
T T
 (1)   (2) 
 y   y 

  
y = Q⋅x
(1) (1)
y ⋅y
(1)
x ⋅ QT ⋅ Q ⋅ x (2)
(2)
(1)
x (1) ⋅ x (2)
= = = cos α
y (2)= Q ⋅ x (2) y (1) y (2) y (1) y (2) x (1) x (2)

 Consequently, the rotation y= Q ⋅ x maintains angles and


distances.

61
Polar Decomposition of F
 Consider the deformation gradient tensor, F:

F = Q ⋅ U = V ⋅Q
stretching
 
rotation
 
dx =F ⋅ dX =( V ⋅ Q ) ⋅ dX =V ⋅ ( Q ⋅ dX )

(not)
F (•) ≡ stretching  rotation (•)

rotation

stretching
 
dx =F ⋅ dX =( Q ⋅ U ) ⋅ dX =Q ⋅ ( U ⋅ dX )
REMARK
For a rigid body motion:
not
F(•) ≡ rotation  stretching (•) U= V= 1 and Q = F

62
2.9 Volume Variation
Ch.2. Deformation and Strain

63
Differential Volume Ratio
 Consider the variation of a differential volume associated to a particle
P:
=
dV0 ( dX
(1) ( 2)
× dX (3)
)=
⋅ dX
 dX 1(1) dX 2(1) dX 3(1) 
deformed  ( 2) ( 2) ( 2)

configuration = det
=  dX 1 dX 2 dX 3  M
 ( 3) ( 3) ( 3) 

dX 1 dX 2

dX 3 

M 
reference  

configuration
=
dVt ( dx( ) × dx( ) )=
1 2
⋅ dx( )
3

 dx1(1) dx2(1) dx3(1) 


 ( 2) 
= det
=  dx1 dx2( 2) dx3( 2)  m
 ( 3) 
 dx1 dx2( 3) dx3( 3) 

 m 

= =
M ij dX (i )
j mij dx (ji )

64
Differential Volume Ratio
 Consider now: dx (i ) =
F ⋅ dX ( i ) i ∈ {1, 2,3} → Fundamental eq. of deformation
 (i )
dx j = Fjk ⋅ dX k(i ) i, j ∈ {1, 2,3}

= =
M ij dX (i )
j and mij dx (ji )

m=
ij dx (j=
i)
Fjk dX k(=
i)
Fjk M ik= M ik FkjT = M ⋅ FT
m

 Then:
dVt = m = M ⋅ FT = M FT = F M = F dV0 dVt = F dV0

dV
0

 And, defining J (X,t) as the jacobian of the deformation,


J ( X, t ) det F ( X, t ) > 0
= dVt= J ⋅ dV0

65
2.10 Area Variation
Ch.2. Deformation and Strain

66
Surface Area Ratio
 Consider the variation of a differential area associated to a particle P:
dA := dAN → material vector "differential of area" → dA = dA
da := dan → spatial vector "differential of area" → da = dA
Deformed (current)
configuration ( 3)
dV0 = dH dA = d
X
 ⋅
N dA =
Reference (initial) dH
configuration =dX(3) ⋅ N
 dA =d A ⋅ d X ( 3)

dA

dVt = dh da = x( ) ⋅ n da =
3
d
dh
=dx ( 3 ) ⋅ n
 da =d a ⋅ d x ( 3)

da

67
Surface Area Ratio

 Consider now:
da ⋅ dx (
3)
dV=
t

dx (3)= F ⋅ dX( )
3

dVt = F dV0
dV= dA ⋅ dX( )
3
0

( 3) ( 3) ( 3)
dVt = F d
A ⋅
dX =
  d a ⋅ F ⋅ dX ∀dX ⇒ F dA = da ⋅ F

F dV dV0 dVt
0

da = F ⋅ dA ⋅ F −1 =
da n F N ⋅ F −1dA da F N ⋅ F −1 dA
=
dA = N dA
da = n da
68
2.11 Volumetric Strain
Ch.2. Deformation and Strain

69
Volumetric Strain
 Volumetric Strain:
def dV ( X, t ) − dV ( X, t0 ) not dVt − dV0
e ( X, t ) =
dV ( X, t ) dV0

dVt = F dV0

F dV0 − dV0
e=
dV0

=
e F −1 REMARK
The incompressibility condition (null
volumetric strain) takes the form
e = J −1 = 0 ⇒ J = F = 1

70
2.12 Infinitesimal Strain
Ch.2. Deformation and Strain

71
Infinitesimal Strain Theory
 The infinitesimal strain theory (also called t0 t
small strain theory) is based on the Ω
simplifying hypotheses: P u P’
Ω0 X
x
 Displacements are very small w.r.t. the
typical dimensions in the continuum
medium,

 As a consequence, u << ( size of Ω0 ) and the reference and deformed configurations are
considered to be practically the same, as are the material
not
and spatial coordinates:
Ω ≅ Ω0 and x = X+u ≅ X ( X, t ) u ( X, t ) ≡ u ( x, t )
U=
xi = X i + ui ≅ X i not
U i ( X, t ) =ui ( X, t ) ≡ ui ( x, t ) i ∈ {1, 2,3}

Displacement gradients are infinitesimal, ∂ui


 << 1, ∀i, j ∈ {1, 2,3}.
∂x j

72
Infinitesimal Strain Theory
 The material and spatial coordinates coincide, x = X + 
u ≅X
≈0
 Even though it is considered that u cannot be neglected when calculating
other properties such as the infinitesimal strain tensor ε.
 There is no difference between the material and spatial
differential operators:

symb
∂ ∂
∇ = ˆ
e = eˆ i =

∂ ∂
i
 X i x i
J (=
 X, t ) U( X, t )= ⊗ ∇ u(x, t )=
⊗ ∇ j(x, t )

 The local an material time derivatives coincide


Γ ( X (x, t ) γ =
 , t ) ≈ Γ= (x, t ) γ ( X, t )
 ≈x
 d γ ∂γ ( X, t ) ∂γ ( x, t )

= = = γ
 dt ∂t ∂t
73
Strain Tensors
 Green-Lagrange strain tensor

E= (
1 T
F F − 1)=
1
( J + JT + JT J )  E ≅
1
2
( J + J T
) =
1
2
( j + jT
)= ε
2 2 

1  ∂ui ∂u j ∂uk ∂uk   Eij ≅ = 1  ∂ui ∂u j 
E=  + +  ε ij  + =  ε ij i, j ∈ {1, 2,3}
ij
2  ∂x j ∂xi ∂xi ∂xi   2  ∂x j ∂xi 
∂uk
<< 1
∂x j
 Euler-Almansi strain tensor 
 e ≅
1
2
( j + jT
) =
1
2
( J + JT ) = ε

e=
1
( 1 − F −T F −1 ) =
1
( j + jT − jT j)   ∂u 
2 2  eij ≅ 1  ∂ui + j =  ε ij i, j ∈ {1, 2,3}
1  ∂ui ∂u j ∂uk ∂uk   2  ∂x j ∂xi 
e=  + − 
2  ∂x j ∂xi ∂xi ∂xi  ∂uk
ij
<< 1
∂x j
 Therefore, the infinitesimal strain tensor is defined as :
 REMARK
1
( ) 1
( ) 1 not

=ε J + J=T
j + j=
T
(u ⊗ ∇ + ∇ ⊗ =
u ) ∇ su
 2 2 2 ε is a symmetrical tensor and its components

 ε ij =1  ∂ui ∂u j  are infinitesimal: | ε ij |<< 1, ∀i, j ∈ {1, 2,3}
 +  i, j ∈ {1, 2,3}
 2  ∂x j ∂xi 
74
Stretch and Unit Elongation
 Stretch in terms of the strain tensors:
1
λT = 1 + 2 
T⋅⋅T
E  λt =
x 1− 2 t ⋅e ⋅ t

Considering that e ≅ E ≅ ε and that it is


infinitesimal, a Taylor linear series expansion up to
first order terms around x = 0 yields:
λ ( x=
) 1+ 2 x λ ( x) =
1
dλ 1− 2 x
≅ λ ( 0) + x=
1+ x

dx x =0
 ≅ λ ( 0) + x=
1+ x
=1 dx x =0

λT ≅ 1 + T ⋅ ε ⋅ T =1 λt ≅ 1 + t ⋅ ε ⋅ t

 But in Infinitesimal Strain Theory, T ≈ t. So the linearized stretch and


unit elongation through a direction given by the unit vector T ≈ t are:
ds ds − dS
λ= ≅ 1+ t ⋅ε ⋅t ≅ 1+ T ⋅ε ⋅ T ε= = λ −1 = t ⋅ ε ⋅ t
dS dS
75
Physical Interpretation of
Infinitesimal Strains
 Consider the components of the infinitesimal strain tensor, ε:
ε xx ε xy ε xz  ε11 ε12 ε13 
 
=ε ε xy ε yy ε yz  ≡ ε12 ε 22 ε 23 
ε xz ε yz ε zz  ε13 ε 23 ε 33 

 For a segment parallel to the x-axis, the stretch and unit elongation are:
λ ≅ 1+ t ⋅ε ⋅t
ε11 ε12 ε13  1 
reference t ⋅ ε ⋅ t [1 0 0] ⋅ ε12 ε 22 ε 23  =
= ⋅ 0  ε11
configuration ε13 ε 23 ε 33  0 

λ1 = λx = 1 + ε11 Stretch in the x-direction


1  ds  ε1 =1 − λ =ε11
    Unit elongation in the
T (1)
≅t (1)
≡  0  dX ≅ dx ≡  0  ε x =1 − λ =ε xx x-direction
0  0
   
76
Physical Interpretation of
Infinitesimal Strains
 Similarly, the stretching and unit elongation of the material in the y-
direction and the z-direction:
λ1 = 1 + ε11 ⇒ ε x = λx − 1 = ε xx
λ2 = 1 + ε 22 ⇒ ε y = λ y − 1 = ε yy
λ3 = 1 + ε 33 ⇒ ε z = λz − 1 = ε zz

 The diagonal components of the infinitesimal strain tensor are the unit
elongations of the material when in the x, y and z-directions.
ε xx ε xy ε xz 
 
ε = ε xy ε yy ε yz 
ε xz ε yz ε zz 

77
Physical Interpretation of
Infinitesimal Strains
 Consider the angle between a segment parallel to the X-axis and a
segment parallel to the Y-axis, the angle is Θ = π2 . XY

 Applying: π 2E
θ ≡ θ xy = − arcsin XY
2 1 + 2 EXX 1 + 2 EYY
E XX = ε xx
E XY = ε xy
EYY = ε yy

π 2ε xy π π
reference
θ xy = − arcsin ≅ − arcsin 2ε xy = − 2ε xy
2 1 + 2ε xx 1 + 2ε yy 2  2
configuration     ≈2ε xy
≈1 ≈1

REMARK
The Taylor linear series expansion of arcsin x yields
( x ) + ... = x + O ( x 2 )
d arcsin
arcsin ( x ) ≅ arcsin ( 0 ) +
dx x =0

78
Physical Interpretation of
Infinitesimal Strains
π
θ xy ≅ − 2ε xy
2

 The increment of the final angle w.r.t. its initial value:


π π π
∆θ xy =
θ xy − ≅ − 2ε xy − =−2ε xy
2 2 2
 Similarly, the increment of the final angle w.r.t. its initial value for
couples of segments oriented in the direction of the coordinate axes:
1 1 1
ε xy =− ∆θ xy ; ε xz =− ∆θ xz ; ε yz =− ∆θ yz
2 2 2
 The non-diagonal components of the infinitesimal strain tensor are equal
to the semi-decrements produced by the deformation of the angles
between segments initially oriented in the x, y and z-directions.

79
Physical Interpretation of
Infinitesimal Strains
 In short,
reference deformed
configuration configuration

ε xx = ε x 1
ε xy =− ∆θ xy
2
ε yy = ε y 1
ε xz =− ∆θ xz
2
ε zz = ε z 1
ε yz =− ∆θ yz
2

80
Engineering Strains
 Using an engineering notation, instead of the scientific notation, the
components of the infinitesimal strain tensor are
REMARK
Positive longitudinal strains indicate
increase in segment length.

Positive angular strains indicate the


corresponding angles decrease with
the deformation process.
Angular
strains Longitudinal
strains

 Because of the symmetry of ε, the tensor can be written as a


6-component infinitesimal strain vector, (Voigt’s notation):
def
ε∈R 6
[ε x , ε y , ε z ,
ε= γ xy , γ xz , γ yz ]T
   
longitudinal angular strains
strains
81
Variation of Angles
 Consider two segments in the reference configuration with the same
origin an angle Θ between them.
θ = Θ + ∆θ
T ⋅ (1 + 2E ) ⋅ T
(1) ( 2)
E=ε
cos θ =
(1) (1) ( 2) ( 2)
1+ 2 T ⋅E ⋅T 1+ 2 T ⋅E ⋅T
T(1) ⋅ [1 + 2ε ] ⋅ T(2)
cos(Θ + ∆θ ) =
1 + 2T(1) ⋅ ε ⋅ T(1) 1 + 2T(2) ⋅ ε ⋅ T(2)
reference deformed
configuration ≈1 ≈1
configuration

cos(Θ + ∆θ ) =
T(1) ⋅ T(2) + 2T(1) ⋅ ε ⋅ T(2)

82
Variation of Angles
cos(Θ + ∆θ=) T(1) ⋅ T(2) + 2T(1) ⋅ ε ⋅ T(2)

 T(1) and T(2) are unit vectors in the directions of the original segments,
therefore, T(1) ⋅ T(2)= T(1) T(2) cos Θ
= cos Θ
T(1) ≈ t (1)
 Also, cos(Θ + ∆θ=) cos Θ ⋅ cos ∆θ − sinΘ ⋅ sin∆=
θ cos Θ − sinΘ ⋅ ∆θ T (2)
≈ t (2)

Θ ≈θ
≈1 ≈ ∆θ
2T(1) ⋅ ε ⋅ T(2) 2t (1) ⋅ ε ⋅ t (2)
θ cos Θ + 2T ⋅ ε ⋅ T
cos Θ − sinΘ ⋅ ∆= (1) (2)
∆θ =− =−
sinΘ sinθ

REMARK
The Taylor linear series expansion of sin x
and cos x yield
( x ) + ... = x + O ( x 2 )
d sin
sin ( x ) ≅ sin ( 0 ) +
dx x =0

( x ) + ... =1 + O ( x 2 )
d cos
cos ( x ) ≅ cos ( 0 ) +
dx x =0

83
Polar Decomposition
 Polar decomposition in finite-strain problems:
left polar
not
 decomposition
=
U FT ⋅ F 
not 
V = F ⋅ FT  ⇒ F = Q ⋅ U = V ⋅Q
Q= F ⋅ U −1 =V −1 ⋅ F  right polar
 decomposition

REMARK
In Infinitesimal Strain Theory
x ≈ X , therefore,= ∂x
F ≈1
∂X

84
Polar Decomposition
 In Infinitesimal Strain Theory:
U = FT F = (1 + J ) ⋅ ( 1 + J ) =
T
1 + J + JT + JT ⋅ J ≈ 1 + J + JT = 1 +
1
2
( J + JT )
<< J =x

U= 1 + ε
infinitesimal strain tensor

Similarly, REMARK
The Taylor linear series expansion of 1 + x
U = (1 + ε ) = 1 − ε = 1 − ( J + J T )
−1 1
and (1 + x ) yield
−1 −1

=x 2

x =1 + x + O ( x 2 )
1
λ ( x ) = 1 + x ≅ λ ( 0) +
=ε dx x =0 2
U −1 = 1 − ε λ ( x ) = (1 + x ) ≅ λ ( 0 ) +
−1 dλ
x =1 − x + O ( x 2 )
dx
infinitesimal x =0

strain tensor
85
Polar Decomposition
(1 + J ) ⋅ 1 − ( J + JT ) =+
1 J − ( J + J T ) − J ⋅ ( J + J T ) =+ ( J − JT )
1 1 1 1
F U −1 =
Q =⋅ 1
 2  2 2 2
<< J = Ω
Q= 1 + Ω
 The infinitesimal rotation tensor Ω is defined: REMARK
The antisymmetric or
 def 1 1 def
Ω = (J − J T )= (u ⊗ ∇ − ∇ ⊗ u)= ∇ a u skew-symmetrical
 2 2 gradient operator is

Ω 1  ∂ui ∂u j  defined as:
=  −  << 1 i, j ∈ {1, 2,3}
 ij 2  ∂x j ∂xi  1
  0 Ω12 −Ω31 
∇ a (•=
) [(•) ⊗ ∇ − ∇ ⊗ (•)]
2
 The diagonal terms of Ω are zero: [ Ω ] =  −Ω12 0 Ω 23 
 Ω31 −Ω 23 0 
 It can be expressed as
an infinitesimal rotation vector θ,  ∂u3 ∂u2 
 ∂x − ∂x 
θ1   −Ω 23   2 3 
REMARK
    1  ∂u1 ∂u3  def 1
θ ≡ θ 2 =  −Ω31 =  − = ∇×u Ω is a skew-symmetric
θ   −Ω  2  ∂x3 ∂x1  2 tensor and its components
 3   12   ∂u2 ∂u1 
 −  are infinitesimal.
 ∂x1 ∂x2 

86
Polar Decomposition
 From any skew-symmetric tensor Ω, it can be extracted a vector θ (axial
vector of Ω) exhibiting the following property:
Ω ⋅r =θ×r ∀r
As a consequence:
 The resulting vector is orthogonal to r.

 If the components of Ω are infinitesimal, then Ω ⋅ r =θ × r is also infinitesimal


 The vector r + Ω ⋅ r = r + θ × r can be seen as the result of applying a (infinitesimal)
rotation (of axial vector θ) on the vector r .

87
Proof of θ×r =
Ω⋅r ∀r

 The result of the dot product of the infinitesimal rotation tensor, Ω, and a
generic vector, r, is exactly the same as the result of the cross product of
the infinitesimal rotation vector, θ, and this same vector.
 0 Ω12 −Ω31  θ1   −Ω 23   r1 
[Ω ] =  −Ω12 0 Ω23  → θ ≡ θ 2  =  −Ω31  ⇒ Ω ⋅ r =θ × r ∀r =  r2 
θ   −Ω  r 
 Ω31 −Ω 23 0   3   12   3
 Proof:
eˆ1 eˆ 2 eˆ 3   eˆ1 eˆ 2 eˆ 3   Ω12 r2 − Ω31r3 
 
θ × r = det θ1 θ 2 θ3  = det  −Ω 23 −Ω12  =
not
−Ω31 −Ω12 r1 + Ω 23 r3 
 r1 r2 r3   Ω r −Ω r 
 r1 r2 r3   31 1 23 2 

 0 Ω12 −Ω31   r1   Ω12 r2 − Ω31r3 


    
Ω ⋅ r =−Ω12
 0 Ω 23  r2 = −Ω12 r1 + Ω 23 r3 
 Ω31 −Ω 23 0   r3  
 Ω31r1 − Ω 23 r2 

88
Polar Decomposition
 Using:
J= F − 1


1
2
( J + JT ) 1 J =+
F =+ 1
1
2
( J + JT ) + ( J − JT )
1
2 F = 1+ε+Ω
Q= 1 + Ω =ε = Ω

 Consider a differential segment dX:


stretch
 rotation

dx = F ⋅ dX = ( 1 + ε + Ω ) ⋅ d X = ε ⋅ d X + ( 1 + Ω ) ⋅ dX

F(•) ≡ stretching (•) + rotation (•)

REMARK
The infinitesimal rotation tensor
characterizes the rotation and, in the
small-strain context, maintains angles
89 and distances.
Volumetric Deformation
 The volumetric strain:
=
e F −1

 Considering: F= Q ⋅ U and U= 1 + ε

1 + ε xx ε xy ε xz
F = Q ⋅ U = Q U = U = 1 + ε = det ε xy 1 + ε yy ε yz =
ε xz ε yz 1 + ε zz
= 1 + ε xx + ε yy + ε zz + O ( ε 2 ) ≈ 1 + Tr ( ε )
= Tr ( ε )

e = Tr ( ε )

90
2.13 Strain Rate
Ch.2. Deformation and Strain

REMARK
We are no longer assuming an
infinitesimal strain framework
91
Spatial Velocity Gradient Tensor
 Consider the relative velocity between two points in space at a given
(current) instant:
∂v
=
 v P′ v= (x, t ) v ( x1 , x2 , x3 , t ) dv = ⋅ dx =⋅
l dx
 ∂
x
dv (x, t ) = v Q′ − v P′ = v ( x + dx, t ) − v ( x, t ) l
∂vi
=
dvi = dx j lij dx j
∂x j

lij i, j ∈ {1, 2,3}

 ( )= v ⊗ ∇
def ∂v x, t

 l ( x , t )=
Spatial velocity  ∂x

gradient tensor = lij ∂vi i, j ∈ {1, 2,3}
 ∂x j

92
Strain Rate and Rotation Rate (or Spin)
Tensors
 The spatial velocity gradient tensor can be split into a symmetrical and
a skew-symmetrical tensor:

l= v ⊗ ∇
 sym [ l ] + skew [ l ] =
l= :d + w
l ∂vi i, j ∈ 1, 2,3
=
 ij ∂x { }
 j

Strain Rate Tensor Rotation Rate or Spin Tensor


( l + l )=
def
1 1 not
d= sym(l)= T
( v ⊗ ∇ + ∇ ⊗ v )= ∇ s v def
w= skew (l)=
1
( l − l )=
T 1 not
( v ⊗ ∇ − ∇ ⊗ v )= ∇ a v
2 2 2 2
1  ∂v ∂v  1  ∂v ∂v 
d ij = i + j  i, j ∈ {1, 2,3} w ij = i − j  i, j ∈ {1, 2,3}
2  ∂x j ∂xi  2  ∂x j ∂xi 
 d11 d12 d 31   0 w12 − w 31 
[d ] = d12 d 22 d 23  [ w ] =  − w12 0 w 23 
d 31 d 23 d 33   w 31 − w 23 0 

93
Physical Interpretation of d
 The strain rate measures the rate of deformation of the square of the
differential length ds in the spatial configuration,
d d d d  dx   dx 
( ds(t =
)) ( dx ⋅ d=x) ( dx ) ⋅ dx + dx ⋅ ( d=
x ) d   ⋅ dx + dx ⋅ d  =  dv ⋅ dx + dx ⋅ dv
2

dt dt dt dt  
dt  
dt
dv = l ⋅ dx =v =v
1
= d (l + lT )
2
d
dt
( ds(t ) )
2
= (
d x ⋅ l ) ⋅ d x + d x ⋅ ( l ⋅ dx ) =
 
T

dx ⋅  lT + l  ⋅ dx = 2dx ⋅ d ⋅ dx
dv 
T dv  = 2d
   

Differentiating w.r.t. time the expression ( ds(t ) ) − ( dS )= 2 dX ⋅ E ⋅ dX


=
2 2

d
dt
( (ds (t )) 2 − (
dS ) 2 ) =
d
dt
( 2dX ⋅ E ( X, t ) ⋅ dX ) = 2dX ⋅
dE
dt
⋅ dX =
d
dt
( (ds (t )) 2 )
 constant
2dX⋅E X,t ⋅dX notation
  
= E

94
Physical Interpretation of d
 ⋅ dX = dx ⋅ d ⋅ dx
dX ⋅ E
dx= F ⋅ dX
 ⋅ dX = dx ⋅ d ⋅ dx =
dX ⋅ E [
dx ] [ d ][ dx ] = [ F ⋅ dX ] [d ][ F ⋅ dX ] =
T
   
T

 
dX ⋅ ( FT ⋅ d ⋅ F) ⋅ dX
F⋅dX F⋅dX T T
dX  F  F dX 
  

 And, rearranging terms:


  ⋅ dX= 0 ∀dX
dX ⋅ FT ⋅ d ⋅ F − E =
FT ⋅ d ⋅ F − E  = FT ⋅ d ⋅ F
E
  0

 There is a direct relation between the material derivative of the material strain
tensor and the strain rate tensor but they are not the same.
  and d will coincide when in the
E
reference configuration F |t =t0 = 1 .
REMARK
Given a 2 order tensor A,
nd

if x ⋅ A ⋅ x =0 for any vector


x ≠ 0 then A = 0 .

95
Physical Interpretation of w
 To determine the (skew-symmetric) rotation rate (spin) tensor only three
different components are needed:
 0 w12 w13 
1  ∂vi ∂v j 
w ij = −  i, j ∈ {1, 2,3} [ w ] =  − w12 0 w23 
2  ∂x j ∂xi   − w13 − w23 0 

 The spin vector (axial vector [w]) of can be extracted:


  ∂v 2 ∂v3  
−  − 
  3 ∂x ∂x2 
   − w23   ω1   0 −ω3 ω2 
1   ∂v3 ∂v1   w=  ω = ω −ω1 
1 1
ω
= rot ( =
v) ∇×v ≡ − − =
    2 [w]  3 0
2   ∂x1 ∂x3  
13
2 2
   − w12  ω3   −ω2 ω1 0 
  ∂v1 ∂v 2  
 −  ∂x − ∂x  
  2 1 

 The vector 2ω = ∇ × v is named vorticity vector.

96
Physical Interpretation of w
 It can be proven that the equality ω× r =w ⋅ r ∀r holds true.
Therefore:
 ω is the angular velocity of a rotation movement.
 ω x r = w · r is the rotation velocity of the point that
has r as its position vector w.r.t. the rotation centre.

 Consider now the relative velocity dv,


dv = l ⋅ dx
l= d + w

dv =d ⋅ dx + w ⋅ dx

97
2.14 Material time Derivatives
Ch.2. Deformation and Strain

99
Deformation Gradient Tensor F
 The material time derivative of the deformation gradient tensor,

∂xi ( X, t ) REMARK
=
Fij ( X, t ) i, j ∈ {1, 2,3} The equality of cross derivatives
∂X j
applies here: ∂ 2 (•) = ∂ 2 (•)
d ∂µi µ j ∂µ j µi
dt
dFij ∂ ∂xi ( X, t ) ∂ ∂xi ( X, t ) ∂Vi ( X, t ) ∂vi ( x ( X, t ) ) ∂x k
= = = = = lik Fkj
dt ∂t ∂X j ∂X j ∂t ∂X j ∂xk ∂X j
=Vi ( X, t ) =lik =Fkj

 dF notation 
 dt = F= l ⋅ F

 dF=ij
F= lik Fkj i, j ∈ {1, 2,3}
 dt ij

100
Inverse Deformation Gradient Tensor F-1
 The material time derivative of the inverse deformation gradient
tensor,
1
F ⋅ F −1 = REMARK
Do not mistake the material derivative
d of the inverse tensor for the inverse of
dt the material derivative of the tensor:
d ( F −1 )
( F ( X,t ) ) ≠ ( F
 ( X,t ) )
d −1 dF −1 d −1
(F ⋅ F ) = ⋅F + F⋅ = 0 −1
dt dt dt dt
d ( F −1 ) dF −1
⇒ F⋅ =− ⋅ F =−F ⋅ F −1
dt dt
Rearranging terms,
d ( F −1 )
−F −1 ⋅ F ⋅ F −1 =
= −F −1 ⋅ l ⋅ F ⋅ F −1 =
−F −1 ⋅ l
dt  d ( F −1 )
= l⋅F =1  = −F −1 ⋅ l
 dt
 −1
 dFij
 dt = − F −1
ik lkj i, j ∈ {1, 2,3}

101
Strain Tensor E
 The material time derivative of the material strain tensor has already
been derived for the physical interpretation of the deformation rate
tensor: 
E= F ⋅d ⋅F
T

 A more direct procedure yields the same result:

=
E
2
( F ⋅ F − 1)
1 T

F = l ⋅ F
d F=T
F T ⋅ lT
dt
dE  1  T
dt
= E=
2
( F ⋅ F + FT ⋅ F ) =
2
( F ⋅ l ⋅ F + FT ⋅ l ⋅ F ) = FT ⋅ ( l + lT ) ⋅ F = FT ⋅ d ⋅ F
1 T T 1
2
 
d

 = FT ⋅ d ⋅ F
E

102
Strain Tensor e
 The material time derivative of the spatial strain tensor,

e=
1
2
( 1 − F −T ⋅ F −1 )
F=−1
F −1 ⋅ l
d F −T= lT ⋅ F −T
dt

− ( F −T ⋅ F −1 + F −T ⋅ F −1 ) =( lT ⋅ F −T ⋅ F −1 + F −T ⋅ F −1 ⋅ l )
de 1 1
e =
=
dt 2 2

e = ( l ⋅ F ⋅ F + F −T ⋅ F −1 ⋅ l )
1 T −T −1
2

103
Volume differential dV
 The material time derivative of the volume differential associated to a
given particle,

dV ( x( X, t ), t ) = F( X, t ) dV0 ( X) The material time derivative of the determinant of the


deformation gradient tensor is:
d
dt For a 2nd order
d ∂ F( X, t ) d tensor A:  d A = d A= A ⋅ A −1
dV ( t ) = dV0 F dV0  
∂t
ji
dt dt  dA ij dAij

d F d F dFij −1 dFij −1
= = F F= F F F lik
d dt dFij dt
ji
dt 
kj ji
( dV=) ( ∇ ⋅ v ) F dV0 lik Fkj ( F⋅F−1 ) δ
= ki
dt
= dV ki

∂v i dF
= F=
lii F = F ∇⋅v = F ∇ ⋅ v = (∇ ⋅ v ) F
d ∂xi
 dt
( dV (x, t ) )= ∇ ⋅ v(x, t ) dV (x, t ) ∇⋅v
dt

104
Area differential vector da
 The material time derivative of the area differential associated to a
given particle,

da ( x( X, t ), t ) = F ( X, t ) ⋅ dA( X) ⋅ F −1 ( X, t ) = F ⋅ dA ⋅ F −1

d
dt

dA ⋅ F −1 + F ⋅ dA ( F −1 )
d dF d
(t )
da=
dt dt dt
= F ∇⋅v = −F −1 ⋅ l

d
( da ) = ( ∇ ⋅ v ) F dA ⋅ F −1 − F dA ⋅ F −1 ⋅ l
dt
= da = da

d
( da ) = da ( ∇ ⋅ v ) − da ⋅ l = da ⋅ 1(∇ ⋅ v) − da ⋅ l = da ⋅ ( (∇ ⋅ v)1 − l )
dt da⋅1

105
2.15 Other Coordinate Systems
Ch.2. Deformation and Strain

106
Curvilinear Orthogonal Coord. System
 A curvilinear coordinate system is defined by:
 The coordinates, generically named {a, b, c}
 Its vector basis, {eˆ a , eˆ b , eˆ c } , formed by unit vectors eˆ=
a eˆ=
b e=
ˆ c 1.

 If the elements of the basis are orthogonal is is called an orthogonal


coordinate system: eˆ a ⋅ eˆ b = eˆ a ⋅ eˆ c = eˆ b ⋅ eˆ c = 0
 The orientation of the curvilinear basis may change at each point in
space, .
eˆ m ≡ eˆ m (x) m ∈ {a, b, c}

REMARK
A curvilinear orthogonal coordinate system can be seen as a mobile Cartesian
coordinate system { x′, y′, z ′} , associated to a curvilinear basis {eˆ a , eˆ b , eˆ c } .

108
Curvilinear Orthogonal Coord. System
 A curvilinear orthogonal coordinate system can be seen as a mobile
Cartesian coordinate system {eˆ a , eˆ b , eˆ c } , associated to a curvilinear basis
{ x′, y′, z′} .
 The components of a vector and a tensor magnitude in the curvilinear
orthogonal basis will correspond to those in the given Cartesian local
system:
 va   v x′   Taa Tab Tac  Tx′x′ Tx′y′ Tx′z′ 
     
v ≡  v b  ≡  v y′  T ≡ Tba Tbb Tbc  ≡  Ty′x′ Ty′y′ Ty′z′ 
     Tca Tcc   Tz′x′ Tz′z′ 
 v c   v z′  Tcb Tz′y′

 The components of the curvilinear operators will not be the same as


those in the given Cartesian local system.
 They must be obtained for each specific case.

109
Cylindrical Coordinate System

← z− coordinate line
← θ − coordinate line  x = r cos θ

x( r , θ , z ) ≡  y = r sin θ
← r− coordinate line z = z

∂eˆ r ∂eˆθ
= eˆθ = −eˆ r
∂θ ∂θ

dV = r dθ dr dz

110
Cylindrical Coordinate System
 Nabla operator ∂ 
 ∂r 
 
∂ 1 ∂ ∂  1 ∂ 
∇= eˆ r + eˆθ + eˆ z ⇒ ∇≡
∂r r ∂θ ∂z  r ∂θ   x = r cos θ
  
∂ x( r , θ , z ) ≡  y = r sin θ
 
 ∂z  z = z

 Displacement vector
ur 
u= u r eˆ r + uθ eˆθ + u z eˆ z ⇒ u=  uθ 
 u z 

 Velocity vector  vr 
v= v r eˆ r + vθ eˆθ + v z eˆ z ⇒ u=  vθ 
 v z 

111
Cylindrical Coordinate System
 Infinitesimal strain tensor
ε x′x′ ε x′y′ ε x′z′   ε rr ε rθ ε rz 
ε=
1
2
{
[u ⊗ ∇ ] + [u ⊗ ∇ ]
T
}  
≡ ε x′y′ ε y′y′ ε y′z′ =

ε
 rθ εθθ εθ z 
ε x′z′ ε y′z′ ε z′z′   ε rz εθ z ε zz 
 
∂u 1  1 ∂u r ∂uθ uθ 
ε rr = r ε= + − 
2  r ∂θ

∂r ∂r r 
1 ∂uθ u r 1  ∂u r ∂u z 
εθθ
= + =ε rz  + 
r ∂θ r 2  ∂ z ∂r 
∂u 1  ∂uθ 1 ∂u z 
ε zz = z εθ z
= +
∂z  
2  ∂z r ∂θ 

 x = r cos θ

x( r , θ , z ) ≡  y = r sin θ
z = z

112
Cylindrical Coordinate System
 Strain rate tensor
 d x′x′ d x′y′ d x′z′   d rr d rθ d rz 
d=
1
2
{
[v ⊗ ∇] + [v ⊗ ∇]
T
} 
≡  d x′y′ d y′y′

d y′z′ =
d
 rθ dθθ dθ z 
 d x′z′ d y′z′ d z′z′   d rz dθ z d zz 

∂v 1  1 ∂v r ∂vθ vθ 
d rr = r d= + − 
2  r ∂θ

∂r ∂r r 
1 ∂vθ v r 1  ∂v r ∂v z 
=
dθθ + =
d rz  + 
r ∂θ r 2  ∂ z ∂r 
∂v 1  ∂vθ 1 ∂v z 
d zz = z = +
∂z dθ z  
2  ∂z r ∂θ 

 x = r cos θ

x( r , θ , z ) ≡  y = r sin θ
z = z

113
Spherical Coordinate System

 x = r sin θ cos φ
=x x ( r ,θ , ϕ ) =

≡  y r sin θ sin φ r− coordinate line →
 z = r cos θ

∂eˆ r ∂eˆθ ∂eˆ φ


=
eˆθ =
−eˆ r =
0
∂θ ∂θ ∂θ

dV = r 2 sin θ dr dθ dφ

114
Spherical Coordinate System
 Nabla operator  ∂ 
 ∂ r 
 
∂ 1 ∂ ∂ 1 ∂ 
eˆ φ ⇒ ∇ ≡   x = r sin θ cos φ
1
∇= eˆ r + eˆθ + 
∂r r ∂θ r sin θ ∂φ r ∂θ  = x x ( r , θ , φ ) ≡=
 y r sin θ sin φ
   z = r cos θ
 1 ∂  
 r sin θ ∂φ 

 Displacement vector ur 


 
u= u r eˆ r + uθ eˆθ + uφ eˆ φ ⇒ u=  uθ 
 uφ 
 

 Velocity vector  vr 
 
v= v r eˆ r + vθ eˆθ + vφ eˆ φ ⇒ u=  vθ 
 vφ 
 

115
Spherical Coordinate System
 Infinitesimal strain tensor
ε x′x′ ε x′y′ ε x′z′   ε rr ε rθ ε rφ 
ε=
1
2
{[u ⊗ ∇ ] + [u ⊗ ∇ ]
T
}  
≡ ε x′y′ ε y′y′ ε y′z′ =


εθ r εθθ εθφ 

ε x′z′ ε y′z′ ε z′z′  ε rφ εθφ ε φφ 
  
∂u r
ε rr =
∂r
1 ∂uθ u r
εθθ
= +  x = r sin θ cos φ
r ∂θ r 
1 ∂uφ uθ u =x x ( r , θ , φ ) ≡=
 y r sin θ sin φ
ε ϕϕ= + cotgφ + r  z = r cos θ
r sin θ ∂φ r r 

1  1 ∂u r ∂uθ uθ 
ε= + − 
2  r ∂θ

∂r r 
1  1 ∂u r ∂uφ uφ 
=ε rφ + − 
2  r sin θ ∂φ ∂r r 
1  1 ∂uθ 1 ∂uφ uφ 
=εθφ + − φ
2  r sin θ ∂φ r ∂θ 
cotg
r 

116
Spherical Coordinate System
 Deformation rate tensor
 d x′x′ d x′y′ d x′z′   d rr d rθ d rφ 
d=
1
2
{[v ⊗ ∇] + [v ⊗ ∇]
T
} 
≡  d x′y′ d y′y′

d y′z′ =

 d rθ dθθ dθφ 

 d x′z′ d y′z′ d z′z′   d rφ dθφ dφφ 
 
∂v r
d rr =
∂r  x = r sin θ cos φ

1 ∂vθ v r =x x ( r , θ , φ ) ≡=
 y r sin θ sin φ
=
dθθ +  z = r cos θ
r ∂θ r 
1 ∂vφ vθ v
dφφ = + cotgϕ + r
r sin θ ∂φ r r
1  1 ∂v r ∂vθ vθ 
d= + − 
2  r ∂θ

∂r r 
1  1 ∂v r ∂vφ vφ 
= + − 
2  r sin θ ∂φ
d rφ
∂r r 
1  1 ∂vθ 1 ∂vφ vφ 
= + − φ
2  r sin θ ∂φ r ∂θ 
dθφ cotg
r 

117
Chapter 2
Strain

rs
ee
s gin
2.1 Introduction

t d le En

r
ba
ge ro or
eS m
ci
f
Definition 2.1. In the broader context, the concept of deformation no

ra
C d P cs
longer refers to the study of the absolute motion of the particles as
b
a
i
seen in Chapter 1, but to the study of the relative motion, with respect
an an n

to a given particle, of the particles in its differential neighborhood.


y ha

le
liv or ec
M

.A

2.2 Deformation Gradient Tensor


m

d
uu

Consider the continuous medium in motion of Figure 2.1. A particle P in the


e
X Th


reference configuration Ω0 occupies the point in space P in the present config-
er
tin

uration Ωt , and a particle Q situated in the differential neighborhood of P has


on

.O

relative positions with respect to this particle in the reference and present times
given by dX and dx, respectively. The equation of motion is given by
C


©

not
x = ϕ (X,t) = x (X,t)
not . (2.1)
xi = ϕi (X1 , X2 , X3 ,t) = xi (X1 , X2 , X3 ,t) i ∈ {1, 2, 3}

Differentiating (2.1) with respect to the material coordinates X results in the



Fundamental ⎪
⎨ dx = F · dX
equation of
⎪ ∂ xi (2.2)
deformation ⎩ dxi = dX j = Fi j dX j i, j ∈ {1, 2, 3}
∂ Xj

41
42 C HAPTER 2. S TRAIN

rs
Figure 2.1: Continuous medium in motion.

ee
s gin
Equation (2.2) defines the material deformation gradient tensor F (X,t) 1 .

t d le En

⎪ not
⎨F = x⊗∇

r
ba
Material deformation

ge ro or
∂ xi

eS m
(2.3)
gradient tensor ⎪
⎩ Fi j = i, j ∈ {1, 2, 3}

ci
∂ Xj f

ra
C d P cs
b
a
i
an an n

The explicit components of tensor F are given by


y ha

⎡ ⎤
∂ x1 ∂ x1 ∂ x1
le
liv or ec

⎡ ⎤ ⎢ ⎥
x  ⎢ ∂X ∂ X2 ∂ X3 ⎥
 ⎢ 1⎥  ∂
M

⎢ 1 ⎥
.A

 ∂ ∂ ⎢ ∂ x2 ∂ x2 ∂ x2 ⎥
[F] = x ⊗ ∇ = ⎢ ⎥
⎣ x2 ⎦ ∂ X1 , ∂ X2 , ∂ X3 = ⎢ ⎥ . (2.4)
⎢ ∂ X1 ∂ X3 ⎥
m

∂ X2
 ⎢ ⎥
d

x3   ⎣ ∂x ∂ x3 ⎦
uu

 T ∂ x3
e

   3
X Th


er

∂ ∂ X2 ∂ X3
tin

[x] X1
on

.O
C

Remark 2.1. The deformation gradient tensor F (X,t) contains the


©

information of the relative motion, along time t, of all the material


particles in the differential neighborhood of a given particle, identi-
fied by its material coordinates X. In effect, equation (2.2) provides
the evolution of the relative position vector dx in terms of the cor-
responding relative position in the reference time, dX. Thus, if the
value of F (X,t) is known, the information associated with the gen-
eral concept of deformation defined in Section 2.1 is also known.

1 Here, the symbolic form of the material Nabla operator, ∇ ≡ ∂ êi /∂ Xi , applied to the
not
expression of the open or tensor product, [a ⊗ b]i j = [a b]i j = ai b j , is considered.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Deformation Gradient Tensor 43

2.2.1 Inverse Deformation Gradient Tensor


Consider now the inverse equation of motion
 not
X = ϕ −1 (x,t) = X (x,t) ,
(2.5)
Xi = ϕi−1 (x1 , x2 , x3 ,t) = Xi (x1 , x2 , x3 ,t)
not
i ∈ {1, 2, 3} .

Differentiating (2.5) with respect to the spatial coordinates xi results in




⎨ dX = F−1 · dx ,

rs
⎪ ∂ Xi (2.6)
⎩ dXi = dx j = Fi−1
j dx j i, j ∈ {1, 2, 3} .

ee
∂xj

s gin
The tensor defined in (2.6) is named spatial deformation gradient tensor or in-
verse (material) deformation gradient tensor and is characterized by2

t d le En

r
ba

ge ro or
not
⎨ F−1 = X ⊗ ∇

eS m
ci
Spatial deformation
∂ Xi f (2.7)

ra
gradient tensor ⎪ −1
⎩ Fi j = i, j ∈ {1, 2, 3}
C d P cs
∂xj
b
a
i
an an n
y ha

le
liv or ec

Remark 2.2. The spatial deformation gradient tensor, denoted in


M

(2.6) and (2.7) as F−1 , is in effect the inverse of the (material) defor-
.A

mation gradient tensor F. The verification is immediate since3


m

∂ xi ∂ Xk ∂ xi not
uu
e

= = δi j =⇒ F · F−1 = 1 ,
X Th

∂ Xk ∂ x j ∂xj
er
tin

 
Fik F −1
on

.O

kj
C

∂ Xi ∂ xk ∂ Xi not
F−1 · F = 1 .
©

= = δi j =⇒
∂ xk ∂ X j ∂ Xj
 
Fik−1 Fk j

2 Here, the symbolic form of the spatial Nabla operator, ∇ ≡ ∂ êi /∂ xi , is considered. Note
the difference in notation between this spatial operator ∇ and the material Nabla ∇.
3 The two-index operator Delta Kronecker δ is defined as δ = 1 if i = j and δ = 0 if
ij ij ij
i = j. The second-order unit tensor 1 is given by [1]i j = δi j .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
44 C HAPTER 2. S TRAIN

The explicit components of tensor F−1 are given by


⎡ ⎤
∂ X1 ∂ X1 ∂ X1
⎡ ⎤ ⎢ ⎥
X1   ⎢ ∂x ∂ x2 ∂ x3 ⎥
 −1  ⎢ ⎥ ∂ ∂ ∂ ⎢ 1 ⎥
⎢ ∂ X2 ∂ X2 ∂ X2 ⎥
F = [X ⊗ ∇] = ⎢ ⎥
⎣ X2 ⎦ ∂ x1 , ∂ x2 , ∂ x3 = ⎢ ⎥ . (2.8)
⎢ ∂ x1 ∂ x2 ∂ x3 ⎥
X3    ⎢⎣ ∂X

∂ X3 ∂X ⎦
   [∇]T
3 3

[X] ∂ x1 ∂ x2 ∂ x3

rs
ee
Example 2.1 – At a given time, the motion of a continuous medium is defined

s gin
by ⎧
⎨ x1 = X1 − AX3

t d le En
x2 = X2 − AX3 .

r

ba
x3 = −AX1 + AX2 + X3

ge ro or
eS m
ci
f
Obtain the material deformation gradient tensor F (X,t) at this time. By

ra
C d P cs
b
a
means of the inverse equation of motion, obtain the spatial deformation gra-
i
dient tensor F−1 (x). Using the results obtained, verify that F · F−1 = 1.
an an n
y ha

le
Solution
liv or ec
M

.A

The material deformation gradient tensor is


⎡ ⎤
m

X1 − AX3
d

 T ⎢  
⎥ ∂
uu

∂ ∂
e

not
F = x ⊗ ∇ ≡ [x] ∇ = ⎣ ⎢ ⎥
X2 − AX3 ⎦ ∂ X1 , ∂ X2 , ∂ X3
X Th

er
tin

−AX1 + AX2 + X3
on

.O

⎡ ⎤
−A
C

1 0
not ⎢ ⎥
©

F≡⎣ 0 1 −A ⎦ .
−A A 1
The inverse equation of motion is obtained directly from the algebraic inver-
sion of the equation of motion,
⎡   ⎤
X1 = 1 + A2 x1 − A2 x2 + Ax3
not ⎢   ⎥
X (x,t) ≡ ⎢ ⎥
⎣ X2 = A x1 + 1 − A x2 + Ax3 ⎦ .
2 2

X3 = Ax1 − Ax2 + x3

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Displacements 45

Then, the spatial deformation gradient tensor is


⎡  ⎤
1 + A2 x1 − A2 x2 + Ax3  
⎢   ⎥ ∂ ∂ ∂
−1 not T ⎢
F = X ⊗ ∇ ≡ [X] [∇] = ⎣ A x1 + 1 − A x2 + Ax3 ⎦
2 2 ⎥ , ,
∂ x1 ∂ x2 ∂ x3
Ax1 − Ax2 + x3
⎡ ⎤
1 + A2 −A2 A
not ⎢ ⎥
F−1 ≡ ⎣ A2 1 − A2 A⎦ .

rs
A −A 1

ee
Finally, it is verified that

s gin
⎡ ⎤⎡ ⎤ ⎡ ⎤
0 −A 1 + A2 −A2

t d le En
1 A 1 0 0
not ⎢ ⎥⎢ ⎥ ⎢ ⎥ not
F · F−1 ≡ ⎣ 0 1 −A ⎦ ⎣ A2 1 − A2 A⎦ = ⎣0 1 0⎦ ≡ 1 .

r
ba
ge ro or
eS m
−A A 1 A −A 1 0 0 1

ci
f

ra
C d P cs
b
a
i
an an n
y ha

2.3 Displacements
le
liv or ec
M

.A

Definition 2.2. A displacement is the difference between the posi-


m

tion vectors in the present and reference configurations of a same


d
uu

particle.
e
X Th

er
tin
on

.O

The displacement of a particle P at a given time is defined by vector u, which



C

joins the points in space P (initial position) and P (position at the present time t)
©

of the particle (see Figure 2.2). The displacement of all the particles in the con-
tinuous medium defines a displacement vector field which, as all properties of
the continuous medium, can be described in material form U (X,t) or in spatial
form u (x,t) as follows.

U (X,t) = x (X,t) − X
(2.9)
Ui (X,t) = xi (X,t) − Xi i ∈ {1, 2, 3}

u (x,t) = x − X (x,t)
(2.10)
ui (x,t) = xi − Xi (x,t) i ∈ {1, 2, 3}

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
46 C HAPTER 2. S TRAIN

rs
ee
s gin
Figure 2.2: Displacement of a particle.

t d le En

r
2.3.1 Material and Spatial Displacement Gradient Tensors

ba
ge ro or
eS m
ci
Differentiation with respect to the material coordinates of the displacement vec-
f

ra
C d P cs
tor Ui defined in (2.9) results in
b
a
i
∂Ui ∂ xi ∂ Xi
an an n

de f
= − = Fi j − δi j = Ji j , (2.11)
y ha

∂ Xj ∂ Xj ∂ Xj
 
le
liv or ec

Fi j δi j
M

.A

which defines the material displacement gradient tensor as follows.


m


uu


e

de f
⎨ J (X,t) = U (X,t) ⊗ ∇ = F − 1
X Th

er

Material displacement
tin

gradient tensor ⎪ ∂Ui (2.12)


⎩ Ji j = = Fi j − δi j i, j ∈ {1, 2, 3}
∂ Xj
on

.O
C



©

⎨ U = J · dX
⎪ ∂Ui (2.13)
⎩ dUi = dX j = Ji j dX j i, j ∈ {1, 2, 3}
∂ Xj
Similarly, differentiation with respect to the spatial coordinates of the expres-
sion of ui given in (2.10) yields
∂ ui ∂ xi ∂ Xi de f
= − = δi j − Fi−1
j = ji j , (2.14)
∂xj ∂xj ∂xj
 
δi j Fi−1
j

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Strain Tensors 47

which defines the spatial displacement gradient tensor as follows.



⎪ de f
⎨ j (x,t) = u (x,t) ⊗ ∇ = 1 − F−1
Spatial displacement
⎪ ∂ ui (2.15)
gradient tensor ⎩ ji j = = δi j − Fi−1 i, j ∈ {1, 2, 3}
∂xj j



⎨ u = j · dx
⎪ ∂ ui (2.16)
⎩ dui = dx j = ji j dx j i, j ∈ {1, 2, 3}

rs
∂xj

ee
s gin
2.4 Strain Tensors

t d le En
Consider now a particle of the continuous medium that occupies the point in

r
space P in the material configuration, and another particle Q√in its differen-

ba
ge ro or
eS m
dS = dX · dX) from

ci
tial neighborhood separated a segment dX (with length

f

ra
the previous paticle, being dx (with length ds = dx · dx) its counterpart in
C d P cs
b
a
the present configuration (see Figure 2.3). Both differential vectors are related
i
an an n

through the deformation gradient tensor F (X,t) by means of equations (2.2) and
y ha

(2.6),
le

liv or ec

⎨ dx = F · dX and dX = F−1 · dx ,
M

.A

(2.17)
⎩ dxi = Fi j dX j and dXi = F −1 dx j i, j ∈ {1, 2, 3} .
m

ij
d
uu
e

Then,
X Th


er
tin

⎨ (ds)2 = dx · dx not
≡ [dx]T [dx] = [F · dX]T [F · dX] ≡ dX · FT · F · dX
not
on

.O

(2.18)
⎩ (ds)2 = dxk dxk = Fki dXi Fk j dX j = dXi Fki Fk j dX j = dXi F T Fk j dX j
ik
C

or, alternatively4 ,
⎧  −1 T  −1 

⎪ (dS) 2
= dX · dX
not
≡ [dX] T
[dX] = F · dx F · dx =



⎨ ≡ dx · F−T · F−1 · dx ,
not

(2.19)

⎪ (dS)2 = dXk dXk = Fki−1 dxi Fk−1 −1 −1

⎪ j dx j = dxi Fki Fk j dx j =


= dxi Fik−T Fk−1
j dx j .

 T not
4 The convention (•)−1 = (•)−T is used.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
48 C HAPTER 2. S TRAIN

rs
ee
s gin
Figure 2.3: Differential segments in a continuous medium.

t d le En

r
2.4.1 Material Strain Tensor (Green-Lagrange Strain Tensor)

ba
ge ro or
eS m
ci
Subtracting expressions (2.18) and (2.19) results in
f

ra
C d P cs
b
a
(ds)2 − (dS)2 = dX · FT · F · dX − dX · dX =
i
an an n

= dX · FT · F · dX − dX · 1 · dX =
y ha

 
le
= dX · FT · F − 1 · dX = 2 dX · E · dX , (2.20)
liv or ec

  
M

.A

de f
= 2E
m

which implicitly defines the material strain tensor or Green-Lagrange strain


uu

tensor as follows.
e
X Th

er


tin

⎪  
Material ⎨ E (X,t) = 1 FT · F − 1
on

.O

2 (2.21)
(Green-Lagrange)
⎪ 1 
strain tensor ⎩ Ei j (X,t) = Fki Fk j − δi j i, j ∈ {1, 2, 3}
C

2
©

Remark 2.3. The material strain tensor E is symmetric. Proof is ob-


tained directly from (2.21), observing that
⎧  
⎨ ET = 1 FT · F − 1T = 1 FT · FT T − 1T = 1 FT · F − 1 = E ,
2 2 2
⎩E = E i, j ∈ {1, 2, 3} .
ij ji

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Strain Tensors 49

2.4.2 Spatial Strain Tensor (Almansi Strain Tensor)


Subtracting expressions (2.18) and (2.19) in an alternative form yields

(ds)2 − (dS)2 = dx · dx − dx · F−T · F−1 · dx =


= dx · 1 · dx − dx · F−T · F−1 · dx =
 
= dx · 1 − F−T · F−1 · dx = 2 dx · e · dx , (2.22)
  
de f
= 2e

rs
which implicitly defines the spatial strain tensor or Almansi strain tensor as

ee
follows.

s gin
⎪  
Spatial ⎨ e (x,t) = 1 1 − F−T · F−1

t d le En
(Almansi) 2   (2.23)

strain tensor ⎩ ei j (x,t) =
1
δi j − Fki−1 Fk−1 i, j ∈ {1, 2, 3}

r
2 j

ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
Remark 2.4. The spatial strain tensor e is symmetric. Proof is ob-
an an n

tained directly from (2.23), observing that


y ha

⎧ 
⎪     
T = 1 1 − F−T · F−1 T = 1 1T − F−1 T · F−T T =
 
le

liv or ec


⎪ e
⎨ 2 2
M

.A

1 −T −1


⎪ = 1−F ·F =e,
⎪ 2
m


⎩e = e
d

ij ji i, j ∈ {1, 2, 3} .
uu
e
X Th

er
tin
on

.O

Example 2.2 – Obtain the material and spatial strain tensors for the motion
C

in Example 2.1.
©

Solution
The material strain tensor is ⎛⎡ ⎤⎡ ⎤ ⎡ ⎤⎞
1 0 −A 1 0 −A 1 0 0
1 T  not 1
E (X,t) = F · F − 1 ≡ ⎝⎣ 0 1 A ⎦⎣ 0 1 −A ⎦−⎣ 0 1 0 ⎦⎠ =
2 2
−A −A 1 −A A 1 001
⎡ 2 ⎤
A −A2 −2A
1
= ⎣ −A2 A2 0 ⎦
2
−2A 0 2A2

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
50 C HAPTER 2. S TRAIN

and the spatial strain tensor is


1 
e (X,t) = 1 − F−T · F−1 =
2
⎛⎡ ⎤ ⎡ ⎤⎡ ⎤⎞
1 0 0 1 + A2 A2 A 1 + A2 −A2 A
not 1
≡ ⎝⎣ 0 1 0 ⎦ − ⎣ −A2 1 − A2 −A ⎦ ⎣ A2 1 − A2 A ⎦⎠ =
2
0 0 1 A A 1 A −A 1
⎡ ⎤
−3A2 − 2A4 A2 + 2A4 −2A − 2A3
1
= ⎣ A2 + 2A4 A2 − 2A4 ⎦.

rs
2A3
2
−2A − 2A −2A

ee
3 2A3 2

s gin
Observe that E = e.

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
Remark 2.5. The material strain tensor E and the spatial strain ten-
b
a
sor e are different tensors. They are not the material and spatial de-
i
an an n

scriptions of a same strain tensor. Expressions (2.20) and (2.22),


y ha

le
(ds)2 − (dS)2 = 2dX · E · dX = 2dx · e · dx ,
liv or ec
M

.A

clearly show this since each tensor is affected by a different vector


(dX and dx, respectively).
m

d
uu

The Green-Lagrange strain tensor is naturally described in mate-


e

rial description (E (X,t)). In equation (2.20) it acts on element dX


X Th

er
tin

(defined in material configuration) and, hence, its denomination as


material strain tensor. However, as all properties of the continuous
on

.O

medium, it may be described, if required, in spatial form (E (x,t))


C

through the adequate substitution of the equation of motion.


©

The contrary occurs with the Almansi strain tensor: it is naturally


described in spatial form and in equation (2.22) acts on the differ-
ential vector dx (defined in the spatial configuration) and, thus, its
denomination as spatial strain tensor. It may also be described, if
required, in material form (e (X,t)).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Variation of Distances: Stretch and Unit Elongation 51

2.4.3 Strain Tensors in terms of the Displacement (Gradients)


Replacing expressions (2.12) and (2.15) into equations (2.21) and (2.23) yields
the expressions of the strain tensors in terms of the material displacement gradi-
ent, J (X,t), and the spatial displacement gradient, j (x,t).

1   1 
E= 1 + JT · (1 + J) − 1 = J + JT + JT · J
2 2
 (2.24)
1 ∂Ui ∂U j ∂Uk ∂Uk
Ei j = + + i, j ∈ {1, 2, 3}
2 ∂ X j ∂ Xi ∂ Xi ∂ X j

rs
ee
1  1

s gin
  
e= 1 − 1 − jT · (1 − j) = j + jT − jT · j
2 2


t d le En
(2.25)
1 ∂ ui ∂ u j ∂ uk ∂ uk
ei j = + − i, j ∈ {1, 2, 3}

r
2 ∂ x j ∂ xi ∂ xi ∂ x j

ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n

2.5 Variation of Distances: Stretch and Unit Elongation


y ha

Consider now a particle P in the reference configuration and another particle


le
liv or ec

Q, belonging to the differential neighborhood of P (see Figure 2.4). The corre-


M

.A

sponding positions in the present configuration are given by the points in space
 
P and Q such that the distance between the two particles in the reference con-
m

figuration, dS, is transformed into ds at the present time. The vectors T and t are
uu

 
e

the unit vectors in the directions PQ and P Q , respectively.


X Th

er
tin
on

.O

Definition 2.3. The stretch or stretch ratio of a material point P (or



C

a spatial point P ) in the material direction T (or spatial direction t )


©

 
is the length of the deformed differential segment P Q per unit of
length of the original differential segment PQ.

The translation of the previous definition into mathematical language is

 
de f PQ ds
Stretch = λT = λt = = (0 < λ < ∞) . (2.26)
PQ dS

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
52 C HAPTER 2. S TRAIN

rs
ee
s gin
Figure 2.4: Differential segments and unit vectors in a continuous medium.

t d le En

r
ba
ge ro or
eS m
ci
f
Definition 2.4. The unit elongation, elongation ratio or extension of

ra

C d P cs
a material point P (or a spatial point P ) in the material direction T
b
a
i
5
(or spatial direction t ) is the increment of length of the deformed
an an n

 
y ha

differential segment P Q per unit of length of the original differen-


le
tial segment PQ.
liv or ec
M

.A
m

The corresponding mathematical definition is


d
uu
e
X Th

Δ PQ ds − dS
er

de f
tin

Unit elongation = εT = εt = = . (2.27)


PQ dS
on

.O
C

Equations (2.26) and (2.27) allow immediately relating the values of the unit
©

elongation and the stretch for a same point and direction as follows.
ds − dS ds
ε= = −1 = λ − 1 (⇒ −1 < ε < ∞) (2.28)
dS dS

λ

5 Often, the subindices (•)T and (•)t will be dropped when referring to stretches or unit
elongations. However, one must bear in mind that both stretches and unit elongations are
always associated with a particular direction.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Variation of Distances: Stretch and Unit Elongation 53

Remark 2.6. The following deformations may take place:


• If λ = 1 (ε = 0) ⇒ ds = dS: The particles P and Q may have
moved along time, but without increasing or decreasing the dis-
tance between them.
• If λ > 1 (ε > 0) ⇒ ds > dS: The distance between the particles
P and Q has lengthened with the deformation of the medium.
• If λ < 1 (ε < 0) ⇒ ds < dS: The distance between the particles

rs
P and Q has shortened with the deformation of the medium.

ee
s gin
t d le En
2.5.1 Stretches, Unit Elongations and Strain Tensors

r
ba
ge ro or
Consider equations (2.21) and (2.22) as well as the geometric expressions

eS m
ci
dX = T dS and dx = t ds (see Figure 2.4). Then,
f

ra
C d P cs

b
a

⎪ (ds)2 − (dS)2 = 2 dX · E · dX = 2 (dS)2 T · E · T
i

an an n


⎨  
y ha

dS T dS T (2.29)
le

⎪ (ds)2 − (dS)2 = 2 dx · e · dx = 2 (ds)2 t · e · t
liv or ec


⎪  

M

.A

ds t ds t
m

and dividing these expressions by (dS)2 and (ds)2 , respectively, results in


d
uu
e


X Th


er
tin

ds 2 λ = 1 + 2T · E · T
−1 = λ −1 = 2 T·E·T ⇒
2
√ (2.30)
dS ε = λ − 1 = 1 + 2T · E · T − 1
on

.O

  
C

λ
©

  1
λ=√
dS 2 1 2
1 − 2t · e · t
1− = 1− = 2 t·e·t ⇒ 1 (2.31)
ds λ ε = λ −1 = √ −1
   1 − 2t · e · t
1/λ

These equations allow calculating the unit elongation and stretch for a given
direction (in material description, T, or in spatial description, t ).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
54 C HAPTER 2. S TRAIN

Remark 2.7. The material and spatial strain tensors, E (X,t) and
e (x,t), contain information on the stretches (and unit elongations)
for any direction in a differential neighborhood of a given particle,
as evidenced by (2.30) and (2.31).

Example 2.3 – The spatial strain tensor for a given motion is

rs
⎡ ⎤

ee
0 0 −tetz
not ⎢ ⎥

s gin
e (x,t) ≡ ⎣ 0 0 0 ⎦.
−tetz t (2etz − et )

t d le En
0

r
Calculate the length, at time t = 0, of the segment that at time t = 2 is recti-

ba
ge ro or
eS m
linear and joins points a ≡ (0, 0, 0) and b ≡ (1, 1, 1).

ci
f

ra
C d P cs
b
a
Solution
i
an an n

The shape and geometric position of the material segment at time t = 2 is


y ha

known. At time t = 0 (reference time) the segment is not necessarily recti-


le
liv or ec

linear and the positions of its extremes A and B (see figure below) are not
M

.A

known. To determine its length, (2.31) is applied for a unit vector in the di-
rection of the spatial configuration t,
m

1 ds 1
uu

λ=√ = =⇒ dS = ds .
e

λ
X Th

1 − 2 t · e · t dS
er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Variation of Angles 55


To obtain the stretch in the direction t ≡ [1, 1, 1]T / 3, the expression t · e · t
not

is computed first as
⎡ ⎤⎡ ⎤
0 0 −tetz 1
not 1 ⎢ ⎥⎢ ⎥ 1 1 t
t·e·t ≡ √ [1, 1, 1] ⎣ 0 0 0 ⎦ ⎣ 1 ⎦ √ = − te .
3 3 3
−tetz 0 t (2etz − et ) 1

Then, the corresponding stretch at time t = 2 is


" √

rs
1 " 1 3
λ=! =⇒ λ" =! =√ .

ee
1 + 23 tet t=2
1+ 3e
4 2 3 + 4e2

s gin
The length at time t = 0 of the segment AB is

t d le En
# B # b # b
1 1 1 1√

r
lAB = dS = ds = ds = lab =

ba
3

ge ro or
λ λ λ λ

eS m
A a a
  

ci
f

ra
lab
C d P cs
b
a
i
and replacing the expression obtained above for the stretch at time t = 2
an an n
y ha

finally results in $
le
lAB = 3 + 4e2 .
liv or ec
M

.A
m

d
uu

2.6 Variation of Angles


e
X Th

er
tin

Consider a particle P and two additional particles Q and R, belonging to the dif-
ferential neighborhood of P in the material configuration (see Figure 2.5), and
on

.O

  
the same particles occupying the spatial positions P , Q and R . The relationship
C

between the angles that form the corresponding differential segments in the ref-
©

erence configuration (angle Θ ) and the present configuration (angle θ ) is to be


considered next.
Applying (2.2) and (2.6) on the differential vectors that separate the particles,
⎧ ⎧
⎨ dx(1) = F · dX(1) ⎨ dX(1) = F−1 · dx(1)
=⇒ (2.32)
⎩ dx(2) = F · dX(2) ⎩ dX(2) = F−1 · dx(2)

and using the definitions of the unit vectors T(1) , T(2) , t(1) and t(2) that establish
the corresponding directions in Figure 2.5,

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
56 C HAPTER 2. S TRAIN

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
Figure 2.5: Angles between particles in a continuous medium.

ci
f

ra
C d P cs
⎧ ⎧
b
a
⎨ dX(1) = dS(1) T(1) ⎨ dx(1) = ds(1) t(1) ,
i
an an n

=⇒ (2.33)
y ha

⎩ dX(2) = dS(2) T(2) ⎩ dx(2) = ds(2) t(2) .


le
liv or ec
M

Finally, according to the definition in (2.26), the corresponding stretches are


.A



m


⎪ 1
d

⎨ ds(1) = λ (1) dS(1) ⎨ dS(1) = ds(1) ,


uu

λ (1)
e

=⇒ (2.34)
X Th

⎩ ds(2) = λ (2) dS(2) ⎪


er


tin

⎩ dS(2) = 1
ds(2) .
λ (2)
on

.O

Expanding now the scalar product6 of the vectors dx(1) and dx(2) ,
C

" "" "  T  


©

" "" "


dS(1) dS(2) cos θ = "dx(1) " "dx(2) " cos θ = dx(1) · dx(2) ≡ dx(1) dx(2) =
not

 T    
= F · dX(1) F · dX(2) ≡ dX(1) · FT · F · dX(2) = dX(1) · (2E + 1) · dX(2)
not

1 1
= dS(1) T(1) · (2E + 1) · T(2) dS(2) = (1) ds(1) T(1) · (2E + 1) · T(2) (2) ds(2) =
λ λ
(1) (2) 1 1 (1) (2)
= ds ds T · (2E + 1) · T ,
λ (1) λ (2)
(2.35)

6 The scalar product of two vectors a and b is defined in terms of the angle between them, θ ,
as a · b = |a| · |b| cos θ .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Physical Interpretation of the Strain Tensors 57

and, comparing the initial and final terms in (2.35), yields

T(1) · (1 + 2E) · T(2)


cos θ = , (2.36)
λ (1) λ (2)
where the stretches λ (1) and λ (2) can be obtained by applying (2.30) to the
directions T(1) and T(2) , resulting in

T(1) · (1 + 2E) · T(2)


cos θ = $ $ . (2.37)
1 + 2T(1) · E · T(1) 1 + 2T(2) · E · T(2)

rs
In an analogous way, operating on the reference configuration, the angle Θ

ee
between the differential segments dX(1) and dX(2) (in terms of t(1) , t(2) and e )

s gin
is obtained,

t d le En
t(1) · (1 − 2e) · t(2)
cosΘ = $ $ .

r
(2.38)

ba
ge ro or
1 − 2t(1) · e · t(1) 1 − 2t(2) · e · t(2)

eS m
ci
f

ra
C d P cs
b
a
i
Remark 2.8. Similarly to the discussion in Remark 2.7, the material
an an n

and spatial strain tensors, E (X,t) and e (x,t), also contain informa-
y ha

tion on the variation of the angles between differential segments in


le
liv or ec

the differential neighborhood of a particle during the deformation


M

process. These facts will be the basis for providing a physical inter-
.A

pretation of the components of the strain tensors in Section 2.7.


m

d
uu
e
X Th

er
tin

2.7 Physical Interpretation of the Strain Tensors


on

.O
C

2.7.1 Material Strain Tensor


©

Consider a segment PQ, oriented parallel to the X1 -axis in the reference config-
uration (see Figure 2.6). Before the deformation takes place, PQ has a known
length dS = dX.
 
The length of P Q is sought. To this aim, consider the material strain tensor
E given by its components,
⎡ ⎤ ⎡ ⎤
EXX EXY EXZ E11 E12 E13
not ⎢ ⎥ ⎢ ⎥
E ≡ ⎣ EXY EYY EY Z ⎦ = ⎣ E12 E22 E23 ⎦ . (2.39)
EXZ EY Z EZZ E13 E23 E33

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
58 C HAPTER 2. S TRAIN

rs
ee
Figure 2.6: Differential segment in the reference configuration.

s gin
t d le En
Consequently,

r
ba
⎡ ⎤⎡ ⎤

ge ro or
eS m
E11 E12 E13 1

ci
⎣ f⎦ ⎣ 0 ⎦ = E11 .

ra
not T
T · E · T ≡ [T] [E] [T] = [1, 0, 0] E12 E22 E23 (2.40)
C d P cs
b
a
E13 E23 E33 0
i
an an n
y ha

The stretch in the material direction X1 is now obtained by replacing


√ the value
le
T · E · T into the expression for stretch (2.30), resulting in λ1 = 1 + 2E11 . In an
liv or ec

analogous manner, the segments oriented in the directions X2 ≡ Y and X3 ≡ Z


M

.A

are considered to obtain the values λ2 and λ3 as follows.


m

√ √ √
d

λ1 = 1 + 2E11 = 1 + 2EXX ⇒ εX = λX − 1 = 1 + 2EXX − 1


uu
e

√ √ √
X Th

er

λ2 = 1 + 2E22 = 1 + 2EYY ⇒ εY = λY − 1 = 1 + 2EYY − 1


tin

(2.41)
√ √ √
λ3 = 1 + 2E33 = 1 + 2EZZ ⇒ εZ = λZ − 1 = 1 + 2EZZ − 1
on

.O
C

Remark 2.9. The components EXX , EYY and EZZ (or E11 , E22 and
E33 ) of the main diagonal of tensor E (denoted longitudinal strains)
contain the information on stretch and unit elongations of the dif-
ferential segments that were initially (in the reference configuration)
oriented in the directions X, Y and Z, respectively.
• If EXX = 0 ⇒ εX = 0 : No unit elongation in direction X.
• If EYY = 0 ⇒ εY = 0 : No unit elongation in direction Y .
• If EZZ = 0 ⇒ εZ = 0 : No unit elongation in direction Z.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Physical Interpretation of the Strain Tensors 59

rs
ee
Figure 2.7: Angles between differential segments in the reference and present configu-

s gin
rations.

t d le En

r
Consider now the angle between segments PQ (parallel to the X1 -axis) and PR

ba
ge ro or
eS m
(parallel to the X2 -axis), where Q and R are two particles in the differential neigh-

ci
f  

ra
borhood of P in the material configuration and P , Q and R are the respective
C d P cs
b
positions in the spatial configuration (see Figure 2.7). If the angle (Θ = π/2)

a
i
between the segments in the reference configuration is known, the angle θ in
an an n
y ha

the present configuration can be determined using (2.37) and taking into ac-
count their orthogonality ( T(1) · T(2) = 0 ) and the equalities T(1) · E · T(1) = E11 ,
le
liv or ec

T(2) · E · T(2) = E22 and T(1) · E · T(2) = E12 . That is,


M

.A

T(1) · (1 + 2E) · T(2)


m

cos θ = $ $
d
uu

1 + 2T(1) · E · T(1) 1 + 2T(2) · E · T(2)


e

(2.42)
X Th

er
tin

2E12
=√ √ ,
1 + 2E11 1 + 2E22
on

.O

which is the same as


C

π 2EXY
θ ≡ θxy = − arcsin √ √ . (2.43)
2 1 + 2EXX 1 + 2EYY
The increment of the final angle with respect to its initial value results in
2EXY
ΔΘXY = θxy − ΘXY = − arcsin √ √ . (2.44)
 1 + 2EXX 1 + 2EYY
π/2

Analogous results are obtained starting from pairs of segments that are ori-
ented in different combinations of the coordinate axes, resulting in

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
60 C HAPTER 2. S TRAIN

2EXY
ΔΘXY = − arcsin √ √
1 + 2EXX 1 + 2EYY
2EXZ
ΔΘXZ = − arcsin √ √ . (2.45)
1 + 2EXX 1 + 2EZZ
2EY Z
ΔΘY Z = − arcsin √ √
1 + 2EYY 1 + 2EZZ

rs
ee
Remark 2.10. The components EXY , EXZ and EY Z (or E12 , E13 and

s gin
E23 ) of the tensor E (denoted angular strains) contain the informa-
tion on variation of the angles between the differential segments that

t d le En
were initially (in the reference configuration) oriented in the direc-
tions X, Y and Z, respectively.

r
ba
ge ro or
eS m
• If EXY = 0 : The deformation does not produce a variation in the

ci
f
angle between the two segments initially oriented in the direc-

ra
C d P cs
tions X and Y .
b
a
i
• If EXZ = 0 : The deformation does not produce a variation in the
an an n
y ha

angle between the two segments initially oriented in the direc-


tions X and Z.
le
liv or ec

• If EY Z = 0 : The deformation does not produce a variation in the


M

.A

angle between the two segments initially oriented in the direc-


tions Y and Z.
m

d
uu
e
X Th

er
tin

The physical interpretation of the components of the material strain tensor is


shown in Figure 2.8 on an elemental parallelepiped in the neighborhood of a
on

.O

particle P with edges oriented in the direction of the coordinate axes.


C

2.7.2 Spatial Strain Tensor


Arguments similar to those of the previous subsection allow interpreting the
spatial components of the strain tensor,
⎡ ⎤ ⎡ ⎤
exx exy exz e11 e12 e13
not ⎢ ⎥ ⎢ ⎥
e ≡ ⎣ exy eyy eyz ⎦ = ⎣ e12 e22 e23 ⎦ . (2.46)
exz eyz ezz e13 e23 e33

The components of the main diagonal (longitudinal strains) can be interpreted


in terms of the stretches and unit elongations of the differential segments ori-

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Physical Interpretation of the Strain Tensors 61

rs
ee
s gin
Figure 2.8: Physical interpretation of the material strain tensor.

t d le En

r
ented in the direction of the coordinate axes in the present configuration,

ba
ge ro or
eS m
ci
f

ra
C d P cs
1 1 1
b
a
λ1 = √ =√ ⇒ εx = √ −1
i
1 − 2e11 1 − 2exx 1 − 2exx
an an n
y ha

1 1 1
λ2 = √ =$ ⇒ εy = $ −1 ,
le
(2.47)
liv or ec

1 − 2e22 1 − 2eyy 1 − 2eyy


M

.A

1 1 1
λ3 = √ =√ ⇒ εz = √ −1
m

1 − 2e33 1 − 2ezz 1 − 2ezz


d
uu
e
X Th

er

while the components outside the main diagonal (angular strains) contain infor-
tin

mation on the variation of the angles between the differential segments oriented
on

.O

in the direction of the coordinate axes in the present configuration,


C

π 2exy
Δ θxy = −ΘXY = − arcsin √ $
2 1 − 2exx 1 − 2eyy
π 2exz .
Δ θxz = −ΘXZ = − arcsin √ √ (2.48)
2 1 − 2exx 1 − 2ezz
π 2eyz
Δ θyz = −ΘY Z = − arcsin $ √
2 1 − 2eyy 1 − 2ezz

Figure 2.9 summarizes the physical interpretation of the components of the spa-
tial strain tensor.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
62 C HAPTER 2. S TRAIN

rs
Figure 2.9: Physical interpretation of the spatial strain tensor.

ee
s gin
2.8 Polar Decomposition

t d le En
The polar decomposition theorem of tensor analysis establishes that, given a

r
ba
second-order tensor F such that |F| > 0, there exist an orthogonal tensor Q 7 and

ge ro or
eS m
ci
two symmetric tensors U and V such that8
f

ra
C d P cs
not √

b
a
U = FT · F ⎪

i

an an n

not √
y ha

V = F·F T =⇒ F = Q · U = V · Q . (2.49)


le

liv or ec

Q = F · U−1 = V−1 · F
M

.A

This decomposition is unique for each tensor F and is denominated left polar
m

decomposition (F = Q · U) or right polar decomposition (F = V · Q). Tensors U


d
uu

and V are named right and left stretch tensors, respectively.


e
X Th

Considering now the deformation gradient tensor and the fundamental re-
er
tin

lation dx = F · dX defined in (2.2) as well as the polar decomposition given


in (2.49), the following is obtained9 .
on

.O

stretching
  
C

rotation
  
©

dx = F · dX = (V · Q) · dX = V · (Q · dX) (2.50)
not
F (•) ≡ stretching ◦ rotation (•)

7 A second-order tensor Q is orthogonal if QT · Q = Q · QT = 1 is verified.


8 To obtain the square root of a tensor, first the tensor must be diagonalized, then the square
root of the elements in the diagonal of the diagonalized component matrix are obtained and,
finally, the diagonalization is undone.
9 The notation (◦) is used here to indicate the composition of two operations ξ and ϕ:
z = ϕ ◦ ξ (x).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Polar Decomposition 63

rotation
  
stretching
  
dx = F · dX = (Q · U) · dX = Q · (U · dX) (2.51)
not
F (•) ≡ rotation ◦ stretching (•)

Remark 2.11. An orthogonal tensor Q (such that |Q| = 1) is named

rs
rotation tensor and the mapping y = Q · x is denominated rotation.

ee
A rotation has the following properties:

s gin
• When applied on any vector x, the result is another vector
y = Q · x with the same modulus,

t d le En

y
2 = y·y ≡ [y]T ·[y] = [Q · x]T ·[Q · x] ≡ x·QT · Q ·x = x·x =
x
2 .
not not

  

r
ba
ge ro or
eS m
1

ci
f

ra
• The result of multiplying (mapping) the orthogonal tensor Q to
C d P cs
b
a
two vectors x(1) and x(2) with the same origin and that form an
i
an an n

angle α between them, maintains the same angle between the


y ha

images y(1) = Q · x(1) and y(2) = Q · x(2) ,


le
liv or ec

y(1) · y(2) x(1) · QT · Q · x(2) x(1) · x(2)


( ( ( ( = ( ( ( ( = ( ( ( ( = cos α .
M

.A

(y(1) ( (y(2) ( (y(1) ( (y(2) ( (x(1) ( (x(2) (


m

d
uu

In consequence, the mapping (rotation) y = Q · x maintains the an-


e
X Th

er

gles and distances.


tin
on

.O
C

Remark 2.12. Equations (2.50) establish that the relative motion in


the neighborhood of the particle during the deformation process
(characterized by tensor F ) can be understood as the composition
of a rotation (characterized by the rotation tensor Q, which main-
tains angles and distances) and a stretching or deformation in itself
(which modifies angles and distances) characterized by the tensor V
(see Figure 2.10).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
64 C HAPTER 2. S TRAIN

rs
ee
s gin
t d le En
Figure 2.10: Polar decomposition.

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n

Remark 2.13. Alternatively, equations (2.51) allow characterizing


y ha

the relative motion in the neighborhood of a particle during the de-


le
formation process as the superposition of a stretching or deformation
liv or ec

in itself (characterized by tensor U ) and a rotation (characterized by


M

.A

the rotation tensor Q ).


m

A rigid body motion is a particular case of deformation characterized


d

by U = V = 1 and Q = F.
uu
e
X Th

er
tin
on

.O

2.9 Volume Variation


C

Consider a particle P of the continuous medium in the reference configuration


(t = 0) which has a differential volume dV0 associated with it (see Figure 2.11).
This differential volume is characterized by the positions of another three par-
ticles Q, R and S belonging to the differential neighborhood of P, which are
aligned with this particle in three arbitrary directions. The volume differential
dVt , associated with the same particle in the present configuration (at time t),
   
will also be characterized by the spatial points P , Q , R and S corresponding
to Figure 2.11 (the positions of which define a parallelepiped that is no longer
oriented along the coordinate axes).
The relative position vectors between the particles in the material configura-
tion are dX(1) , dX(2) and dX(3) , and their counterparts in the spatial configura-

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Volume Variation 65

tion are dx(1) = F · dX(1) , dx(2) = F · dX(2) and dx(3) = F · dX(3) . Obviously, the
relations 
dx(i) = F · dX(i)
(i) (i) (2.52)
dx j = Fjk dXk i, j, k ∈ {1, 2, 3}
are satisfied. Then, the volumes10 associated with a particle in both configura-
tions can be written as
⎡ (1) (1) (1)

dX1 dX2 dX3
  ⎢ (2) ⎥
dV0 = dX(1) × dX(2) · dX(3) = det ⎢ (2) (2) ⎥
⎣ dX1 dX2 dX3 ⎦ = |M| ,

rs
ee
(3) (3) (3)
dX1 dX2 dX3
  

s gin
[M]

t d le En
⎡ (1) (1) (1)

r
dx1 dx2 dx3
 

ba
ge ro or
⎢ (2) ⎥

eS m
dVt = dx(1) × dx(2) · dx(3) = det ⎢ (2) (2) ⎥
⎣ dx1 dx2 dx3 ⎦ = |m| ,

ci
(2.53)
f

ra
C d P cs
(3) (3) (3)

b
a
dx1 dx2 dx3
  
i
an an n

[m]
y ha

le
liv or ec

(i) (i)
where Mi j = dX j and mi j = dx j . Considering these expressions,
M

.A

(i) (i)
mi j = dx j = Fjk dXk = Fjk dMik = dMik FkTj =⇒ m = M · FT (2.54)
m

d
uu
e

is deduced and, consequently11 ,


X Th

er
tin

" " " " ⎫


dVt = |m| = "M · FT " = |M| "FT " = |F| |M| = |F| dV0 ⎪ ⎪

on

.O


dV0 =⇒ dVt = |F|t dV0

C


dVt = dV (x (X,t) ,t) = |F (X,t)| dV0 (X, 0) = |F|t dV0 ⎭
©

(2.55)

10 The volume of a parallelepiped is calculated as the scalar triple product (a × b) · c of the


concurrent edge-vectors a, b and c, which meet at any of the parallelepiped’s vertices. Note
that the scalar triple product is the determinant of the matrix constituted by the components
of the above mentioned vectors arranged "in rows.
"
11 The expressions |A · B| = |A| · |B| and "AT " = |A| are used here.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
66 C HAPTER 2. S TRAIN

rs
ee
s gin
Figure 2.11: Variation of a volume differential element.

t d le En

r
2.10 Area Variation

ba
ge ro or
eS m
ci
f
Consider an area differential dA associated with a particle P in the reference

ra
C d P cs
configuration and its variation along time. To define this area differential, con-
b
a
i
sider two particles Q and R in the differential neighborhood of P, whose relative
an an n

positions with respect to this particle are dX(1) and dX(2) , respectively (see Fig-
y ha

ure 2.12). Consider also an arbitrary auxiliary particle S whose relative position
le
liv or ec

vector is dX(3) . An area differential vector dA = dA N associated with the scalar


M

.A

differential area, dA, is defined. The module of vector dA is dA and its direction
is the same as that of the unit normal vector in the material configuration N.
m

In the present configuration, at time t, the particle will occupy a point in


d
uu


e

space P and will have an area differential da associated with it which, in turn,
X Th

defines an area differential vector da = da n, where n is the corresponding unit


er
tin

normal vector in the spatial configuration. Consider also the positions of the
  
other particles Q , R and S and their relative position vectors dx(1) , dx(2) and
on

.O

dx(3) .
C

The volumes dV0 and dVt of the corresponding parallelepipeds can be calcu-
©

lated as
dV0 = dH dA = dX(3) · N dA = dX(3) · N dA = dA · dX(3)
   
dH dA (2.56)
(3)
dVt = dh da = dx · n da = dx · n da = da · dx(3)
(3)
   
dh da

and, taking into account that dx(3) = F · dX(3) , as well as the expression for
change in volume (2.55), results in
da · F · dX(3) = da · dx(3) = dVt = |F| dV0 = |F| dA · dX(3) ∀dX(3) . (2.57)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Infinitesimal Strain 67

rs
ee
Figure 2.12: Variation of an area differential.

s gin
t d le En
Comparing the first and last terms12 in (2.57) and considering that the relative

r
position of particle S can take any value (as can, therefore, vector dX(3) ), finally

ba
ge ro or
eS m
ci
yields
f

ra
da · F = |F| dA =⇒ da = |F| dA · F−1 .
C d P cs
(2.58)
b
a
i
an an n

To obtain the relation between the two area differential scalars, dA and da,
y ha

expressions dA = N dA and da = n da are replaced into (2.58) and the modules


le
are taken, resulting in
liv or ec

( (
M

.A

da n = |F| N · dF−1 dA =⇒ da = |F| (N · dF−1 ( dA . (2.59)


m

2.11 Infinitesimal Strain


uu
e
X Th

er
tin

Infinitesimal strain theory (also denominated small deformation theory) is based


on two simplifying hypotheses of the general theory (or finite strain theory) seen
on

.O

in the previous sections (see Figure 2.13).


C

Definition 2.5. The simplifying hypotheses are:


1) Displacements are very small compared to the typical dimensions
in the continuous medium:
u
<<
X
.
2) Displacement gradients are very small (infinitesimal).

12 Here, the following tensor algebra theorem is taken into account: given two vectors a and
b, if the relation a · x = b · x is satisfied for all values of x, then a = b.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
68 C HAPTER 2. S TRAIN

rs
ee
s gin
Figure 2.13: Infinitesimal strain in the continuous medium.

t d le En

r
ba
ge ro or
eS m
In accordance with the first hypothesis, the reference configuration Ω0 and

ci
f
the present configuration Ωt are very close together and are considered to be

ra
C d P cs
b
indistinguishable from one another. Consequently, the material and spatial co-

a
i
ordinates coincide and discriminating between material and spatial descriptions
an an n
y ha

no longer makes sense.


⎧ ⎧
le
liv or ec

⎨x = X+u ∼ = X ⎨ U (X,t) not


= u (X,t) ≡ u (x,t)
=⇒
M

.A

⎩ xi = Xi + ui ∼
= Xi ⎩ Ui (X,t) not
= ui (X,t) ≡ ui (x,t) i ∈ {1, 2, 3}
m

(2.60)
d
uu

The second hypothesis can be written in mathematical form as


e
X Th

" "
er
tin

" ∂ ui "
" " ∀ i, j ∈ {1, 2, 3} .
"∂xj " 1 (2.61)
on

.O
C

2.11.1 Strain Tensors. Infinitesimal Strain Tensor


The material and spatial displacement gradient tensors coincide. Indeed, in view
of (2.60),

⎨x = X ∂ ui ∂Ui
j j
=⇒ ji j = = = Ji j =⇒ j = J (2.62)
⎩ ui (x,t) = Ui (X,t) ∂xj ∂ Xj

and the material strain tensor results in

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Infinitesimal Strain 69



⎪ 1  
T + JT · J ∼ 1 J + JT ,


⎪ E = J + J =

⎨ 2 2
 
1 ∂ ui ∂ u j ∂ uk ∂ uk ∼ 1 ∂ ui ∂ u j (2.63)

⎪ Ei j = + + = + ,

⎪ 2 ∂ x j ∂ xi ∂ xi ∂ x j 2 ∂ x j ∂ xi

⎩   
1
where the infinitesimal character of the second-order term (∂ uk ∂ uk /∂ x j ∂ xi ) has
been taken into account. Operating in a similar manner with the spatial strain
tensor,

rs

⎪ 1    
T − jT · j ∼ 1 j + jT = 1 J + JT ,


ee

⎪ e = j + j =
⎨ 2

2 2


s gin
1 ∂ ui ∂ u j ∂ uk ∂ uk ∼ 1 ∂ ui ∂ u j (2.64)

⎪ e ij = + − = + .

⎪ ∂ ∂ ∂ ∂ ∂ ∂

t d le En
⎪ 2 x j x i x i x j 2 x j xi
⎩   
1

r
ba
ge ro or
eS m
ci
Equations (2.63) and (2.64) allow defining the infinitesimal strain tensor (or
f

ra
small strain tensor) ε as13
C d P cs
b
a

i
an an n


⎪ 1  not
y ha


⎨ε = J + JT = ∇s u
le
Infinitesimal 2

liv or ec

(2.65)
strain tensor ⎪
⎪ 1 ∂ ui ∂ u j
⎪ ε
⎩ ij = + i, j ∈ {1, 2, 3}
M

.A

2 ∂ x j ∂ xi
m

d
uu
e
X Th

er
tin

Remark 2.14. Under the infinitesimal strain hypothesis, the material


and spatial strain tensors coincide and collapse into the infinitesimal
on

.O

strain tensor.
E (x,t) = e (x,t) = ε (x,t)
C

Remark 2.15. The infinitesimal strain tensor is symmetric, as ob-


served in its definition in (2.65).
1 T 1  T 
εT = J + JT = J +J = ε
2 2

13 The symmetric gradient operator ∇s is defined as ∇s (•) = ((•) ⊗ ∇ + ∇ ⊗ (•)) /2.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
70 C HAPTER 2. S TRAIN

Remark 2.16. The components of the infinitesimal strain tensor ε are


infinitesimal (εi j 1). Proof is obvious from (2.65) and the condi-
tion that the components of J = j are infinitesimal (see (2.61)).

Example 2.4 – Determine under which conditions the motion in Example 2.1
constitutes an infinitesimal strain case and obtain the infinitesimal strain ten-

rs
sor for this case. Compare it with the result obtained from the spatial and

ee
material strain tensors in Example 2.2 taking into account the infinitesimal
strain hypotheses.

s gin
t d le En
Solution

r
The equation of motion is given by

ba
ge ro or
eS m

ci
⎨ x1 = X1 − AX3
⎪ f

ra
C d P cs
b
a
x2 = X2 − AX3 ,

i

an an n

x3 = −AX1 + AX2 + X3
y ha

le
liv or ec

from which the displacement field is obtained


M

⎡ ⎤
.A

U1 = −AX3
not ⎢ ⎥
m

U (X,t) = x − X ≡ ⎣ U2 = −AX3 ⎦.
d
uu
e

U3 = −AX1 + AX2
X Th

er
tin

It is obvious that, for the displacements to be infinitesimal, A must be in-


on

.O

finitesimal (A 1). Now, to obtain the infinitesimal strain tensor, first the
displacement gradient tensor J (X,t) = j (x,t) must be computed,
C

⎡ ⎤ ⎡ ⎤
−AX3   0 0 −A
not ⎢ ⎥ ∂ ∂ ∂ ⎢ ⎥
J = U⊗∇ ≡ ⎢ ⎣ −AX3 ⎥
⎦ ∂ X1 , ∂ X2 , ∂ X3 = ⎣ 0 0 −A ⎦ .
−AX1 + AX2 −A A 0

Then, the infinitesimal strain tensor, in accordance to (2.65), is


⎡ ⎤
0 0 −A
not ⎢ ⎥
ε = ∇s U ≡ ⎣ 0 0 0 ⎦ .
−A 0 0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Infinitesimal Strain 71

The material and spatial strain tensors obtained in Example 2.2 are, respec-
tively,
⎡ 2 ⎤
A −A2 −2A
not 1 ⎢ ⎥
E (X,t) ≡ ⎣ −A2 A2 0 ⎦ and
2
−2A 0 2A2
⎡ ⎤
−3A2 − 2A4 A2 + 2A4 −2A − 2A3
not 1 ⎢ ⎥
e (X,t) ≡ ⎣ A2 + 2A4 A2 − 2A4 2A3 ⎦.
2
−2A − 2A 3 3 −2A 2

rs
2A

ee
Neglecting
 4 the second-order
 and higher-order infinitesimal terms
A A3 A2 A results in

s gin
⎡ ⎤ ⎡ ⎤

t d le En
0 0 −A 0 0 −A
not ⎢ ⎥ not ⎢ ⎥
0 −A ⎦ =⇒ E = e = ε ,

r
E≡⎣ 0 0 −A ⎦ and e≡⎣ 0

ba
ge ro or
eS m
−A A −A A

ci
0 0
f

ra
C d P cs
b
a
which is in accordance with Remark 2.14.
i
an an n
y ha

2.11.2 Stretch. Unit Elongation


le
liv or ec

Considering
 the general expression
√  (2.30) of the unit elongation in the direction
M

.A

T∼ = t λt = 1 + 2t · E · t and applying a Taylor series expansion14 around 0


(taking into account that E = ε is infinitesimal and, therefore, so is x = t · ε · t ),
m

yields
uu
e


X Th

er

1 + 2t · ε · t ∼
tin

λt = = 1+t·ε ·t
   (2.66)
on

x
.O

εt = λt − 1 = t · ε · t
C

2.11.3 Physical Interpretation of the Infinitesimal Strains


Consider the infinitesimal strain tensor ε and its components in the coordinate
system x1 ≡ x, x2 ≡ y, x3 ≡ z, shown in Figure 2.14,
⎡ ⎤ ⎡ ⎤
εxx εxy εxz ε11 ε12 ε13
not ⎢ ⎥ ⎢ ⎥
ε ≡ ⎣ εxy εyy εyz ⎦ = ⎣ ε12 ε22 ε23 ⎦ . (2.67)
εxz εyz εzz ε13 ε23 ε33
√ √  
14 The Taylor series expansion of 1 + x around x = 0 is 1 + x = 1 + x/2 + O x2 .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
72 C HAPTER 2. S TRAIN

Figure 2.14: Physical interpretation of the infinitesimal strains.

rs
ee
Consider a differential segment PQ oriented in the reference configuration

s gin
parallel to the coordinate axis x1 ≡ x. The stretch λx and the unit elongation εx
in this direction are, according to (2.66) with t = [1, 0, 0]T ,

t d le En
λx = 1 + t · ε · t = 1 + εxx =⇒ εx = λx − 1 = εxx .

r
(2.68)

ba
ge ro or
eS m
ci
This allows assigning to the component εxx ≡ ε11 the physical meaning of unit
f

ra
elongation εx in the direction of the coordinate axis x1 ≡ x. A similar interpre-
C d P cs
b
a
tation is deduced for the other components in the main diagonal of the tensor
i
an an n

ε (εxx , εyy , εzz ),


y ha

εxx = εx ; εyy = εy ; εzz = εz . (2.69)


le
liv or ec

Given now the components outside the main diagonal of ε , consider the dif-
M

.A

ferential segments PQ and PR oriented in the reference configuration parallel to


m

the coordinate directions x and y, respectively. Then, these two segments form
d

an angle Θxy = π/2 in this configuration. Applying (2.43), the increment in the
uu
e

corresponding angle results in15


X Th

er
tin

π εxy ∼
Δ θxy = θxy − = −2 arcsin $ $ = −2 arcsin εxy = −2εxy ,
on

.O

2 1 + 2εxx 1 + 2εyy   
       εxy
C

1 1
©

(2.70)
where the infinitesimal character of εxx , εyy and εxy has been taken into account.
Consequently, εxy can be interpreted from (2.70) as minus the semi-increment,
produced by the strain, of the angle between the two differential segments ini-
tially oriented parallel to the coordinate directions x and y. A similar interpre-
tation is deduced for the other components εxz and εyz ,

1 1 1
εxy = − Δ θxy ; εxz = − Δ θxz ; εyz = − Δ θyz . (2.71)
2 2 2
 
15 The Taylor series expansion of arcsin x around x = 0 is arcsin x = x + O x2 .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Infinitesimal Strain 73

2.11.4 Engineering Strains. Vector of Engineering Strains


There is a strong tradition in engineering to use a particular denomination for
the components of the infinitesimal strain tensor. This convention is named en-
gineering notation, as opposed to the scientific notation generally used in con-
tinuum mechanics. Both notations are synthesized as follows.

engineering notation
 ⎡  
scientific notation
  ⎤
⎡ ⎤ ⎡ ⎤
ε11 ε12 ε13 εxx εxy εxz εx 2 γxy 2 γxz
1 1

⎢ ⎥ ⎢ ⎥ ⎢ ⎥
ε22 ε23 ⎦ ≡ ⎣ εxy εyy εyz ⎦ ≡ ⎢ ⎥
not
ε ≡ ⎣ ε12

rs
⎣2
1
γxy ε y
1
2 γyz ⎦ (2.72)

ee
ε13 ε23 ε33 εxz εyz εzz
2 γxz 2 γyz εz
1 1

s gin
t d le En

r
Remark 2.17. The components in the main diagonal of the strain ten-

ba
ge ro or
eS m
sor (named longitudinal strains) are denoted by ε(•) and coincide

ci
f

ra
with the unit elongations in the directions
 of the  coordinate axes.
C d P cs
Positive values of longitudinal strains ε(•) > 0 correspond to an
b
a
i
increase in length of the corresponding differential segments in the
an an n
y ha

reference configuration.
le
liv or ec
M

.A
m

Remark 2.18. The components outside the main diagonal of the


uu

strain tensor are characterized by the values γ(•, •) (named angu-


e
X Th

er

lar strains) and can be interpreted as the decrements of the corre-


tin

sponding angles oriented in the Cartesian directions


 of the reference
configuration. Positive values of angular strains γ(•, •) > 0 indicate
on

.O

that the corresponding angles close with the deformation process.


C

In engineering, it is also frequent to exploit the symmetry of the infinitesimal


strain tensor (see Remark 2.15) to work only with the six components of the
tensor that are different, grouping them in the vector of engineering strains,
which is defined as follows.

de f
 T
ε ∈ R6 ε = εx , εy , εx , γxy , γxz , γyz
      (2.73)
longitudinal angular
strains strains

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
74 C HAPTER 2. S TRAIN

2.11.5 Variation of the Angle between Two Differential Segments in


Infinitesimal Strain
Consider any two differential segments, PQ and PR, in the reference configu-
ration and the angle Θ they define (see Figure 2.15). The angle formed by the
corresponding deformed segments in the present configuration is θ = Θ + Δ θ .
Applying (2.42) to this case results in

T(1) · (1 + 2εε ) · T(2)


cos θ = cos (Θ + Δ θ ) = $ $ , (2.74)
1 + 2T(1) · ε · T(1) 1 + 2T(2) · ε · T(2)
     

rs
1 1

ee
where T(1) and T(2) are the unit vectors

s gin
( ( (in the( directions of PQ and PR and,
(1) (2) ( (1) ( ( (2) (
therefore, the relation T · T = (T ( (T ( cosΘ = cosΘ is fulfilled. Con-

t d le En
sidering the infinitesimal character of the components of ε and Δ θ , the follow-
ing holds true16 .

r
ba
ge ro or
eS m
ci
f
cos θ = cos (Θ + Δ θ ) = cosΘ · cos Δ θ − sinΘ · sin Δ θ =

ra
     
C d P cs
b
a
≈1 ≈ Δθ
i
an an n
y ha

= cosΘ
  
le
liv or ec

T(1) · T(2) +2T(1) · ε · T(2) (2.75)


= cosΘ − sinΘ · Δ θ = $ $ =
M

.A

1 + T(1) · ε · T(1) 1 + T(2) · ε · T(2)


     
m

≈1 ≈1
d
uu

= cosΘ + 2T(1) · ε · T(2)


e
X Th

er
tin

Therefore, sinΘ · Δ θ = −2T(1) · ε · T(2) and


on

.O

2T(1) · ε · T(2) 2t(1) · ε · t(2)


C

Δθ = − =− , (2.76)
©

sinΘ sin θ

where the infinitesimal character of the strain has been taken into account and,
thus, it follows that T(1) ≈ t(1) , T(2) ≈ t(2) and Θ ≈ θ .

2.11.6 Polar Decomposition


The polar decomposition of the deformation gradient tensor F is given by (2.49)
for the general case of finite strain. In the case of infinitesimal strain, recall-
 
16The following Taylor series expansions around x = 0 are considered: sin x = x + O x2
 2
and cos x = 1 + O x

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Infinitesimal Strain 75

rs
ee
Figure 2.15: Variation of the angle between two differential segments in infinitesimal
strain.

s gin
t d le En
ing (2.12) and the infinitesimal character of the components of the tensor J

r
(see (2.61)), tensor U in (2.49) can be written as17

ba
ge ro or
eS m
$ $ ⎫

ci
U = FT · F = (1 + JT ) · (1 + J) = f ⎪

ra

C d P cs


b
a
$ $  
i
1
= 1 + J + J + J · J ≈ 1 + J + J = 1 + J + J ⎪ =⇒ U = 1 + ε .
an an n

T T T T
   ⎪
y ha

   ⎪
2

J ε
le
liv or ec

(2.77)
M

.A

In a similar manner, due to the infinitesimal character of the components of the


tensor ε (see Remark 2.16), the inverse of tensor U results in18
m

1 
uu

U−1 = (1 + ε )−1 = 1 − ε = 1 −
e

J + JT . (2.78)
X Th

2
er
tin

Therefore, the rotation tensor Q in (2.49) can be written as


on

.O

 ⎫
1  ⎪
C

−1
Q = F · U = (1 + J) · 1 − J + J T = ⎪



©

2 ⎬
1  1   1   =⇒ Q = 1 + Ω .
= 1 + J − J + JT − J · J + JT = 1 + J − JT ⎪ ⎪

2 2   2   ⎪ ⎪

J Ω
(2.79)

√ √  
17 The Taylor series expansions of tensor 1 + x around x = 0 is 1 + x = 1 + x/2 + O x2 .
18 The Taylor series expansions of tensor (1 + x)−1 around x = 0 is (1 + x)−1 = 1 − x +
 
O x2 .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
76 C HAPTER 2. S TRAIN

Equation (2.79) defines the infinitesimal rotation tensor Ω 19 as follows.




⎪ de f 1   1

⎨Ω =
de f
J − JT = (u ⊗ ∇ − ∇ ⊗ u) = ∇a u
Infinitesimal 2 2
 (2.80)
rotation tensor ⎪
⎪ 1 ∂ ui ∂ u j

⎩ Ωi j = − 1 i, j ∈ {1, 2, 3}
2 ∂ x j ∂ xi

Remark 2.19. The tensor Ω is antisymmetric. Indeed,

rs

ee
⎨ Ω T = 1 J − JT T = 1 JT − J = −Ω Ω

s gin
2 2 .
⎩ Ω = −Ω i, j ∈ {1, 2, 3}
ji ij

t d le En
Consequently, the terms in the main diagonal of Ω are zero, and its

r
ba
ge ro or
eS m
matrix of components has the structure

ci
⎡ f ⎤

ra
C d P cs
0 Ω12 −Ω31
b
a
Ω] = ⎣ −Ω12 Ω23 ⎦ .
i
[Ω 0
an an n
y ha

Ω31 −Ω23 0
le
liv or ec
M

.A

In a small rotation context, tensor Ω characterizes the rotation (Q = 1 + Ω )


m

and, thus, the denomination of infinitesimal rotation tensor. Since it is an anti-


uu

symmetric tensor, it is defined solely by three different components (Ω23 , Ω31 ,


e
X Th

Ω12 ), which form the infinitesimal rotation vector θ 20 ,


er
tin

⎡ ⎤
on

.O

Infinitesimal ∂ u 3 ∂ u2
rotation vector: ⎡ ⎤ ⎡ ⎤ ⎢ − ⎥
C

θ1 −Ω23 ⎢ ∂ x2 ∂ x3 ⎥
⎢ ⎥ 1
©

not ⎢ ⎥ ⎢ ⎥ 1⎢ ⎥ de f
θ ≡ ⎣ θ2 ⎦ = ⎣ −Ω31 ⎦ = ⎢ ∂ u1 − ∂ u3 ⎥ = ∇ × u . (2.81)
2 ⎢ ∂ x3 ∂ x1 ⎥ 2
θ3 −Ω12 ⎢ ⎥
⎣ ∂u ∂ u1 ⎦
2

∂ x1 ∂ x2

19 The antisymmetric gradient operator ∇a is defined as ∇a (•) = [(•) ⊗ ∇ − ∇ ⊗ (•)] /2.


20 The operator rotational of (•) is denoted as ∇ × (•).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Infinitesimal Strain 77

Expressions (2.12), (2.65) and (2.79) allow writing


1  1 
F = 1+J+ J + JT + J − JT =⇒ F = 1 + ε + Ω . (2.82)
2 2
     
ε Ω

Remark 2.20. The results of performing a dot product of the in-


finitesimal rotation tensor Ω and performing a cross product of the
infinitesimal rotation vector θ with any vector r ≡ [r1 , r2 , r3 ]T (see

rs
Figure 2.16) coincide. Indeed,

ee
⎡ ⎤⎡ ⎤ ⎡ ⎤

s gin
0 Ω12 −Ω31 r1 Ω12 r2 − Ω31 r3
not ⎢ ⎥⎢ ⎥ ⎢ ⎥
Ω · r ≡ ⎣ −Ω12 0 Ω23 ⎦ ⎣ r2 ⎦ = ⎣ −Ω12 r1 + Ω23 r3 ⎦ ,

t d le En
Ω31 −Ω23 0 r3 Ω31 r1 − Ω23 r2

r
ba
ge ro or
⎡ ⎤ ⎡ ⎤

eS m
ci
ê1 ê2 ê3 ê1 ê2 ê3
f

ra
not ⎢ ⎥ ⎢ ⎥
C d P cs
θ × r ≡ ⎣ θ1 θ2 θ3 ⎦ = ⎣ −Ω23 −Ω31 −Ω12 ⎦ =
b
a
i
an an n

r1 r2 r3 r1 r2 r3
⎡ ⎤
y ha

Ω12 r2 − Ω31 r3
le
⎢ ⎥
liv or ec

= ⎣ −Ω12 r1 + Ω23 r3 ⎦ .
M

.A

Ω31 r1 − Ω23 r2
m

Consequently, vector Ω · r = θ × r has the following characteristics:


uu
e
X Th

• It is orthogonal to vector r (because it is the result of a vector


er
tin

product in which r is involved).


on

.O

• Its module is infinitesimal (because θ is infinitesimal).


C

• Vector r + Ω · r = r + θ × r can be considered, except for higher-


©

order infinitesimals, as the result of applying a rotation θ on


vector r.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
78 C HAPTER 2. S TRAIN

rs
ee
Figure 2.16: Product of the infinitesimal rotation vector and tensor on a vector r.

s gin
t d le En
Consider now a differential segment dX in the neighborhood of a particle P
in the reference configuration (see Figure 2.17). In accordance with (2.82), the

r
ba
ge ro or
stretching transforms this vector into vector dx as follows.

eS m
ci
f

ra
stretching rotation
C d P cs
     
b
a
dx = F · dX = (1 + ε + Ω) · dX = ε · dX + (1 + Ω) · dX
i
an an n

(2.83)
y ha

le
F (•) ≡ stretching (•) + rotation (•)
liv or ec
M

.A
m

d
uu

Remark 2.21. Under infinitesimal strain hypotheses, the expression


e
X Th

in (2.83) characterizes the relative motion of a particle, in the differ-


er
tin

ential neighborhood of this particle, as the following sum:


on

.O

a) A stretching or deformation in itself, characterized by the in-


finitesimal strain tensor ε .
C

b) A rotation characterized by the infinitesimal rotation tensor Ω


which, in the infinitesimal strain context, maintains angles and
distances.
The superposition (stretching ◦ rotation) of the general finite strain
case (see Remark 2.12) degenerates, for the infinitesimal strain case,
into a simple addition (stretching + rotation).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Volumetric Strain 79

Figure 2.17: Stretching and rotation in infinitesimal strain.

rs
ee
s gin
t d le En
2.12 Volumetric Strain

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
Definition 2.6. The volumetric strain is the increment produced by
b
a
i
the deformation of the volume associated with a particle, per unit of
an an n

volume in the reference configuration.


y ha

le
liv or ec
M

.A

This definition can be mathematically expressed as (see Figure 2.18)


m

de f dV (X,t) − dV (X, 0) not dVt − dV0


uu

Volumetric strain: e (X,t) = = . (2.84)


e

dV (X, 0) dV0
X Th

er
tin
on

.O
C

Figure 2.18: Volumetric strain.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
80 C HAPTER 2. S TRAIN

Equation (2.55) allows expressing, in turn, the volumetric strain as follows:


• Finite strain

dVt − dV0 |F|t dV0 − dV0


e= = =⇒ e = |F| − 1 (2.85)
dV0 dV0

• Infinitesimal strain

Considering (2.49) and recalling that Q is an orthogonal tensor (|Q| = 1), yields

rs
⎡ ⎤

ee
1 + εxx εxy εxz
⎢ ⎥

s gin
|F| = |Q · U| = |Q| |U| = |U| = |1 + ε | = det ⎣ εxy 1 + εyy εyz ⎦ ,
εxz εyz 1 + εzz

t d le En
(2.86)

r
ba
where (2.77) has been considered. Taking into account that the components of ε

ge ro or
eS m
ci
are infinitesimal, and neglecting in the expression of its determinant the second-
f

ra
order and higher-order infinitesimal terms, results in
C d P cs
b
a
⎡ ⎤
i
an an n

1 + εxx εxy εxz


 
y ha

⎢ ⎥
|F| = det ⎣ εxy 1 + εyy εyz ⎦ = 1 + εxx + εyy + εzz +O ε 2 ≈ 1 + Tr (εε ) .
le
  
liv or ec

εxz εyz 1 + εzz Tr (εε )


M

.A

(2.87)
m

Then, introducing (2.87) into (2.85) yields, for the infinitesimal strain case
d


uu
e

dVt = (1 + Tr (εε )) dV0 ⎪⎬


X Th

er
tin

dVt − dV0 =⇒ e = Tr (εε ) . (2.88)


e= = |F| − 1 ⎪

on

.O

dV0
C

2.13 Strain Rate


In the previous sections of this chapter, the concept of strain has been studied,
defined as the variation of the relative position (angles and distances) of the
particles in the neighborhood of a given particle. In the following sections, the
rate at which this relative position changes will be considered by introducing
the concept of strain rate as a measure of the variation in the relative position
between particles per unit of time.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Strain Rate 81

2.13.1 Velocity Gradient Tensor


Consider the configuration corresponding to a time t, two particles of the con-
tinuous medium P and Q that occupy the spatial points P and Q at said instant
of time (see Figure 2.19), their velocities vP = v (x,t) and vQ = v (x + dx,t), and
their relative velocity,
dv (x,t) = vQ − vP = v (x + dx,t) − v (x,t) . (2.89)

Then, ⎧

⎪ ∂v

rs
⎨ dv = · dx = l · dx
∂x

ee
, (2.90)


⎪ ∂ vi

s gin

⎩ dvi = dx j = li j dx j i, j ∈ {1, 2, 3}
∂xj

t d le En
where the spatial velocity gradient tensor l (x,t) has been introduced.

r
ba

ge ro or
eS m
⎪ de f ∂ v (x,t)

ci
⎪ l (x,t) =

⎪ f

ra

⎨ ∂x
C d P cs
b
a
Spatial velocity
l = v⊗∇
i
(2.91)
gradient tensor ⎪
an an n


⎪ ∂ vi
y ha


⎪ i, j ∈ {1, 2, 3}
⎩ li j =
∂xj
le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 2.19: Velocities of two particles in the continuous medium.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
82 C HAPTER 2. S TRAIN

2.13.2 Strain Rate and Spin Tensors


The velocity gradient tensor can be split into a symmetric and an antisymmetric
part21 ,
l = d+w , (2.92)
where d is a symmetric tensor denominated strain rate tensor,


⎪ de f 1  1

⎪ d = sym (ll ) = l + l T not
= (v ⊗ ∇ + ∇ ⊗ v) = ∇s v

⎪ 2 2

⎪ 

⎨ d = 1 ∂ vi + ∂ v j
Strain ⎪

rs
ij i, j ∈ {1, 2, 3}
2 ∂ x j ∂ xi

ee
rate (2.93)

tensor ⎪


⎡ ⎤

s gin

⎪ d 11 d 12 d 13



⎪ [d] = ⎣ d12 d22 d23 ⎦

t d le En
d13 d23 d33

r
ba
ge ro or
eS m
ci
and w is an antisymmetric tensor denominated rotation rate tensor or spin ten-
f

ra
sor, whose expression is
C d P cs
b
a
i

an an n


⎪ 1  1
y ha

de f

⎪ w = skew (l
l ) = l − l T not
= (v ⊗ ∇ − ∇ ⊗ v) = ∇a v


le
⎪ 2 2

liv or ec

Rotation⎪

⎨ w = 1 ∂ vi − ∂ v j

i, j ∈ {1, 2, 3}
M

.A

rate ij
2 ∂ x j ∂ xi (2.94)
(spin) ⎪
⎪ ⎡ ⎤
tensor ⎪
m


⎪ 0 w12 −w31
d



uu


⎪ [w] = ⎣ −w12 0 w23 ⎦
e



X Th

er

w31 −w23 0
tin
on

.O
C

2.13.3 Physical Interpretation of the Strain Rate Tensor


©

Consider a differential segment defined by the particles P and Q of Figure 2.20


and the variation of their squared length along time,
d 2 d d d
ds = (dx · dx) = (dx) · dx + dx · (dx) =
dt dt dt  dt
dx dx
=d · dx + dx · d = dv · dx + dx · dv , (2.95)
dt dt
21 Every second-order tensor a can be decomposed into the sum of its symmetric part
(sym (a)) and its antisymmetric or skew-symmetric part (skew (a)) in the form:
a = sym (a) + skew (a) with sym (a) = (a + aT )/2 and skew (a) = (a − aT )/2.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Strain Rate 83

and using relations (2.90) and (2.93), the expression


d 2    
ds = dx · l T · dx + dx · (ll · dx) = dx · l T + l · dx = 2 dx · d · dx (2.96)
dt
is obtained. Differentiating now (2.20) with respect to time and taking into ac-
count (2.96) yields
d d  2 
2 dx · d · dx = ds2 (t) = ds (t) − dS2 =
dt dt (2.97)
d dE .
= (2 dX · E (X,t) · dX) = 2 dX · · dX = 2 dX · E · dX .

rs
dt dt

ee
Replacing (2.2) into (2.97) results in22

s gin
.  
dX · E · dX = dx · d · dx ≡ [dx]T [d] [dx] = [dX]T FT · d · F [dX]
not

t d le En
 . .
=⇒ dX · FT · d · F − E · dX = 0 ∀ dX =⇒ FT · d · F − E = 0

r
ba
ge ro or
eS m
.

ci
E = FT · d · F .
f (2.98)

ra
C d P cs
b
a
i
an an n
y ha

Remark 2.22. Equation (2.98) shows the existing relationship be-


tween the strain rate tensor d (x,t) and the material derivative of
le
liv or ec

.
the material strain tensor E (X,t), providing a physical interpreta-
M

.A

tion (and justifying the denomination) of tensor d (x,t). However,


.
the same equation reveals that tensors d (x,t) and E (X,t) are not
m

exactly the same. Both tensors will coincide in the following cases:
uu

"
e

• In the reference configuration: t = t0 ⇒ F"


X Th

= 1.
er
tin

t=t0
∂x
on

• In infinitesimal strain theory: x ≈ X ⇒ F = ≈ 1.


.O

∂X
C

22 Here, the following tensor algebra theorem is used: given a second-order tensor A, if
x · A · x = 0 is verified for all vectors x = 0, then A ≡ 0.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
84 C HAPTER 2. S TRAIN

Figure 2.20: Differential segment between particles of the continuous medium along

rs
time.

ee
s gin
2.13.4 Physical Interpretation of the Rotation Rate Tensor

t d le En
Taking into account the antisymmetric character of w (which implies it can be

r
defined using only three different components), the vector

ba
ge ro or
eS m
⎡  ⎤

ci
∂ v 2 ∂ v3 f

ra
⎢−
C d P cs
− ⎥ ⎡ ⎤
b
a
⎢ ⎥
⎢  ∂ x3 ∂ x2 ⎥ −w23
i
an an n

not 1 ⎢ ∂ v 3 ∂ v1 ⎥
1 1
ω = rot (v) = ∇ × v ≡ ⎢ ⎥=⎢ ⎥
y ha

⎢ − − ⎥ ⎣ −w31 ⎦ (2.99)
2 2 2⎢ ∂ x1 ∂ x3 ⎥
le
⎢  ⎥ −w12
liv or ec

⎣ ∂ v 1 ∂ v2 ⎦
− −
M

.A

∂ x2 ∂ x1
m

ω = ∇ × v is named vorticiy vector


is extracted from (2.94). Vector 2ω 23 . It can
uu
e

be proven (in an analogous manner to Remark 2.20) that the equality


X Th

er
tin

ω ×r = w·r ∀r (2.100)
on

.O

is satisfied. Therefore, it is possible to characterize ω as the angular velocity of a


C

rotation motion, and ω × r = w · r as the rotation velocity of the point that has r
©

as the position vector with respect to the rotation center (see Figure 2.21). Then,
considering (2.90) and (2.92),
dv = l · dx = (d + w) · dx = d · dx + w · dx , (2.101)
     
stretch rotation
velocity velocity

which allows describing the relative velocity dv of the particles in the neigh-
borhood of a given particle P (see Figure 2.22) as the sum of a relative stretch
23 Observe the similarity in the structure of tensors Ω and θ in Section 2.11.6 and of tensors
w and ω seen here.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Material Time Derivatives of Strain and Other Magnitude Tensors 85

rs
ee
Figure 2.21: Vorticity vector.

s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A

Figure 2.22: Stretch and rotation velocities.


m

d
uu
e

velocity (characterized by the strain rate tensor d) and a relative rotation velocity
X Th

ω ).
er

(characterized by the spin tensor w or the vorticity vector 2ω


tin
on

.O

2.14 Material Time Derivatives of Strain and Other


C

Magnitude Tensors
©

2.14.1 Deformation Gradient Tensor and its Inverse Tensor


Differentiating the expression of F in (2.3) with respect to time24 ,
∂ xi (X,t) dFi j ∂ ∂ xi (X,t) ∂ ∂ xi (X,t)
Fi j = =⇒ = = =⇒ (2.102)
∂ Xj dt ∂t ∂ X j ∂ Xj ∂t
  
vi

24 The Schwartz Theorem (equality of mixed partial derivatives) guarantees that


for a function Φ (x1 , x2 ... xn ) that is continuous and has continuous derivatives,
∂ 2 Φ/(∂ xi ∂ x j ) = ∂ 2 Φ/(∂ x j ∂ xi ) ∀i, j is satisfied.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
86 C HAPTER 2. S TRAIN

dFi j ∂ vi (X,t) ∂ vi (x (X,t)) ∂ xk


= = = lik Fk j =⇒
dt ∂ Xj ∂ xk ∂X
   j
lik Fk j

dF not .
= F = l ·F
dt (2.102 (cont.))
dFi j .
= Fi j = lik Fk j i, j ∈ {1, 2, 3}

rs
dt

ee
s gin
where (2.91) has been taken into account for the velocity gradient tensor l . To
obtain the material time derivative of tensor F−1 , the time derivative of the iden-
tity F · F−1 = 1 is performed25 .

t d le En
 

r
d   dF −1 d F−1

ba
ge ro or
−1 −1
F · F = 1 =⇒ F·F = ·F +F· =0

eS m
ci
 −1  dt dt dt
f

ra
d F .
C d P cs
=⇒ = −F−1 · F · F−1 = −F−1 · l · F · F−1 = −F−1 · l =⇒
b
a
dt    
i
an an n

l ·F 1
y ha

 −1 
le
liv or ec

d F
= −F−1 · l
M

dt
.A

(2.103)
dFi−1
= Fik−1 lk j
j
m

i, j ∈ {1, 2, 3}
d

dt
uu
e
X Th

er
tin

2.14.2 Material and Spatial Strain Tensors


on

.O

From (2.21), (2.102) and (2.93), it follows26


C

1 T  . 1 .T .
©

dE
E= F · F − 1 =⇒ =E= F · F + FT · F =
2 dt 2
1 T T  1  
= F · l · F + FT · l · F = FT · l + l T · F = FT · d · F
2 2   
.
=⇒ E = F · d · F .
T 2d (2.104)
 
25 The material time derivative of the inverse tensor d F−1 /dt must not be confused with
 . −1
the inverse of the material derivative of the tensor: F . These two tensors are completely
different tensors.
26 Observe that the result is the same as the one obtained in (2.98) using an alternative pro-
cedure.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Material Time Derivatives of Strain and Other Magnitude Tensors 87

Using (2.23) and (2.103) for the spatial strain tensor e yields

1 −T −1
 de . 1 d  −T  −1 d  −1 
e= 1−F ·F ⇒ =e=− F · F + F−T · F =
2 dt 2 dt dt
1  T −T −1 
= l · F · F + F−T · F−1 · l
2
. 1  T −T −1 
=⇒ e = l · F · F + F−T · F−1 · l . (2.105)
2

rs
2.14.3 Volume and Area Differentials

ee
The volume differential dV (X,t) associated with a certain particle P varies

s gin
along time (see Figure 2.23) and, in consequence, it makes sense to calculate

t d le En
its material derivative. Differentiating (2.55) for a volume differential results in

r
d d |F|

ba
ge ro or
dV (X,t) = |F (X,t)| dV0 (X) =⇒ dV (t) = dV0 .

eS m
(2.106)

ci
dt dt
f

ra
C d P cs
Therefore, the material derivative of the determinant of the deformation gradient
b
a
tensor |F| is27
i
an an n
y ha

d |F| d |F| dFi j dFi j


= |F| Fji−1 = |F| Fji−1 lik Fk j = |F| Fk j Fji−1 lik =
le
=
liv or ec

dt dFi j dt dt
   
M

.A

lik Fk j [ F·F−1 ] =δ
ki ki
m

∂ vi d |F|
uu

= |F| δki lik = |F| lii = |F| = |F| ∇ · v =⇒ = |F| ∇ · v , (2.107)


e

∂ xi dt
X Th

er
tin

where (2.102) and (2.91) have been considered. Introducing (2.107) into (2.106)
on

.O

and taking into account (2.55) finally results in


C

d
©

(dV ) = (∇ · v) |F| dV0 = (∇ · v) dV . (2.108)


dt
Operating in a similar manner yields the material derivative of the area dif-
ferential associated with a certain particle P and a given direction n (see Fig-
ure 2.24). The area differential vector associated with a particle in the reference
configuration, dA (X) = dA N, and in the present configuration, da (x,t) = da n,
are related through da = |F| · dA · F−1 (see (2.59)) and, differentiating this ex-

27 The derivative of the determinant of a tensor A with respect to the same tensor can be
written in compact notation as d |A|/dA = |A| · A−T or, in index notation, as d |A|/dAi j =
|A| · A−1
ji .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
88 C HAPTER 2. S TRAIN

rs
Figure 2.23: Variation of the volume differential.

ee
s gin
pression, results in

t d le En
d d   d |F| d  −1 
(da) = |F| · dA · F−1 = dA · F−1 + |F| · dA =

r
F

ba
ge ro or
dt dt dt 
  dt  

eS m
ci
|F| ∇ · v −F−1 · l
f

ra
C d P cs
b
a
= (∇ · v) |F| dA · F−1 − |F| dA · F−1 · l =⇒
i
     
an an n
y ha

da da
le
liv or ec

d
(da) = (∇ · v) da − da · l = da · ((∇ · v) 1 − l ) , (2.109)
M

.A

dt
m

where (2.103) and (2.107) have been considered.


uu
e
X Th

er
tin
on

.O
C

Figure 2.24: Variation of the area differential.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Motion and Strains in Cylindrical and Spherical Coordinates 89

2.15 Motion and Strains in Cylindrical and Spherical


Coordinates
The expressions and equations obtained in intrinsic or compact notation are in-
dependent of the coordinate system considered. However, the expressions of the
components depend on the coordinate system used. In addition to the Cartesian
coordinate system, which has been used in the previous sections, two orthogonal
curvilinear coordinate systems will be considered here: cylindrical coordinates
and spherical coordinates.

rs
Remark 2.23. An orthogonal curvilinear coordinate system (gener-

ee
ically referred to as {a, b, c}), is characterized by its physical unit

s gin
basis {êa , êb , êc } (
êa
=
êb
=
êc
= 1), whose components are
orthogonal to each other (êa · êb = êa · êc = êb · êc = 0), as is also the

t d le En
case in a Cartesian system. The fundamental difference is that the

r
orientation of the curvilinear basis changes at each point in space

ba
ge ro or
(êm ≡ êm (x) m ∈ {a, b, c}). Therefore, for the purposes here, an

eS m
ci
f
orthogonal curvilinear coordinate system can be considered as a mo-

ra
bile Cartesian coordinate system {x , y , z } associated with a curvi-
C d P cs
b
a
linear basis {êa , êb , êc } (see Figure 2.25).
i
an an n
y ha

le
liv or ec
M

.A

Remark 2.24. The components, of a certain magnitude of vectorial


m

character (v) or tensorial character (T) in an orthogonal curvilinear


d

coordinate system {a, b, c}, can be obtained as the corresponding


uu
e

components in the local Cartesian system {x , y , z }:


X Th

er
tin

⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
va vx  Taa Tab Tac Tx x Tx y Tx z
on

.O

not ⎢ ⎥ ⎢ ⎥ not ⎢ ⎥ ⎢ ⎥
v ≡ ⎣ vb ⎦ ≡ ⎣ vy ⎦ T ≡ ⎣ Tba Tbb Tbc ⎦ ≡ ⎣ Ty x Ty y Ty z ⎦
C

vc vz Tca Tcb Tcc Tz x Tz y Tz z

Remark 2.25. The curvilinear components of the differential opera-


tors (the ∇ operator and its derivatives) are not the same as their
counterparts in the local coordinate system {x , y , z }. They must be
defined specifically for each case. Their value for cylindrical and
spherical coordinates is provided in the corresponding section.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
90 C HAPTER 2. S TRAIN

2.15.1 Cylindrical Coordinates


The position of a certain point in space can be defined by its cylindrical coordi-
nates {r, θ , z} (see Figure 2.25). The figure also shows the physical orthonormal
basis {êr , êθ , êz }. This basis changes at each point in space according to
∂ êr ∂ êθ
= êθ and = −êr . (2.110)
∂θ ∂θ
Figure 2.26 shows the corresponding differential element. The expressions in
cylindrical coordinates of some of the elements treated in this chapter are:

rs
• Nabla operator, ∇

ee
 T

s gin
∂ 1 ∂ ∂ ∂
not 1 ∂ ∂
∇ = êr + êθ + êz =⇒ ∇≡ , , (2.111)
∂r r ∂θ ∂z ∂r r ∂θ ∂z

t d le En

r
ba
ge ro or
⎡ ⎤

eS m
x = r cos θ

ci
f x (r, θ , z) ≡ ⎣ y = r sin θ ⎦
not

ra
C d P cs
b z=z

a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin

Figure 2.25: Cylindrical coordinates.


on

.O
C

Figure 2.26: Differential element in cylindrical coordinates.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Motion and Strains in Cylindrical and Spherical Coordinates 91

• Displacement vector, u, and velocity vector, v

u ≡ [ur , uθ , uz ]T
not
u = ur êr + uθ êθ + uz êz =⇒ (2.112)

v ≡ [vr , vθ , vz ]T
not
v = vr êr + vθ êθ + vz êz =⇒ (2.113)

• Infinitesimal strain tensor, ε


⎡ ⎤ ⎡ ⎤
εx x εx y εx z εrr εrθ εrz
1 

rs
not ⎢ ⎥ ⎢ ⎥
ε= (u ⊗ ∇) + (u ⊗ ∇)T ≡ ⎣ εx y εy y εy z ⎦ = ⎣ εrθ εθ θ εθ z ⎦

ee
2
εx z εy z εz z εrz εθ z εzz

s gin
∂ ur 1 ∂ uθ ur ∂ uz

t d le En
εrr = εθ θ = + εzz =
∂r r ∂θ r ∂z

r
 

ba
ge ro or
1 1 ∂ ur ∂ uθ uθ ∂ ur ∂ uz

eS m
1
εrθ = + − εrz = +

ci
2 r ∂θ ∂r f ∂z ∂r

ra
r 2
C d P cs

b
a
1 ∂ u θ 1 ∂ uz
i
an an n

εθ z = + (2.114)
2 ∂z r ∂θ
y ha

le
liv or ec

The components of ε are presented on the corresponding differential element in


M

.A

Figure (2.26).
m

• Strain rate tensor, d


uu
e
X Th

⎡ ⎤ ⎡ ⎤
er
tin

dx x dx y dx z drr drθ drz


1 
not ⎢ ⎥ ⎢ ⎥
d= (v ⊗ ∇) + (v ⊗ ∇)T ≡ ⎣ dx y dy y dy z ⎦ = ⎣ drθ dθ θ dθ z ⎦
on

.O

2
C

dx z dy z dz z drz dθ z dzz


©

∂ vr 1 ∂ vθ vr ∂ vz
drr = dθ θ = + dzz =
∂r r ∂θ r ∂z
 
1 1 ∂ vr ∂ vθ vθ 1 ∂ v r ∂ vz
drθ = + − drz = +
2 r ∂θ ∂r r 2 ∂z ∂r

1 ∂ vθ 1 ∂ vz
dθ z = + (2.115)
2 ∂z r ∂θ

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
92 C HAPTER 2. S TRAIN

2.15.2 Spherical Coordinates


A point in space is)defined by * its spherical coordinates {r, θ , φ }. The physical
orthonormal basis êr , êθ , êφ is presented in Figure 2.27. This basis changes at
each point in space according to
∂ êr ∂ êθ ∂ êφ
= êθ , = −êr and =0. (2.116)
∂θ ∂θ ∂θ
The expressions in spherical coordinates of some of the elements treated in this
chapter are:

rs
• Nabla operator, ∇

ee
 

s gin
∂ 1 ∂ 1 ∂ ∂not 1 ∂ 1 ∂ T
∇ = êr + êθ + êφ =⇒ ∇≡ , ,
∂r r ∂θ r sin θ ∂ φ ∂ r r ∂ θ r sin θ ∂ φ

t d le En
(2.117)

r
ba
ge ro or
eS m
ci
• Displacement vector, u, and velocity vector, v f

ra
C d P cs
b
a
not  T
i
u = ur êr + uθ êθ + uφ êφ =⇒ u ≡ ur , uθ , uφ
an an n

(2.118)
y ha

not  T
le
v = vr êr + vθ êθ + vφ êφ =⇒ v ≡ vr , vθ , vφ (2.119)
liv or ec
M

.A
m

d
uu
e
X Th

er
tin

⎡ ⎤
x = r sin θ cos φ
on

.O

x (r, θ , φ ) ≡ ⎣ y = r sin θ sin φ ⎦


not
C

z = z cos θ
©

Figure 2.27: Spherical coordinates.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Motion and Strains in Cylindrical and Spherical Coordinates 93

• Infinitesimal strain tensor, ε


⎡ ⎤ ⎡ ⎤
  εx  x  ε x  y  ε x  z  εrr εrθ εrφ
1 not ⎢ ⎥ ⎢ ⎥
ε= (u ⊗ ∇) + (u ⊗ ∇)T ≡ ⎣ εx y εy y εy z ⎦ = ⎣ εrθ εθ θ εθ φ ⎦
2
εx z εy z εz z εrφ εθ φ εφ φ

∂ ur 1 ∂ uθ ur
εrr = εθ θ = +
∂r r ∂θ r
1 ∂ uφ uθ ur

rs
εφ φ = + cot φ +
r sin θ ∂ φ r r

ee
 
1 1 ∂ ur ∂ uθ uθ 1 ∂ ur ∂ uφ uφ

s gin
1
εrθ = + − εrφ = + −
2 r ∂θ ∂r r 2 r sin θ ∂ φ ∂r r

t d le En

1 1 ∂ uθ 1 ∂ uφ uφ

r
εθ φ = + − cot φ (2.120)

ba
ge ro or
2 r sin θ ∂ φ r ∂θ r

eS m
ci
f

ra
The components of ε are presented on the corresponding differential element in
C d P cs
b
a
Figure 2.28.
i
an an n
y ha

le
• Strain rate tensor, d
liv or ec

⎡ ⎤ ⎡ ⎤
M

.A

  dx x dx y dx z drr drθ drφ


1 not ⎢ ⎥ ⎢ ⎥
d= (v ⊗ ∇) + (v ⊗ ∇)T ≡ ⎣ dx y dy y dy z ⎦ = ⎣ drθ dθ θ dθ φ ⎦
m

2
uu

dx z dy z dz z drφ dθ φ dφ φ


e
X Th

er
tin

∂ vr 1 ∂ vθ vr
drr = dθ θ = +
on

∂r r ∂θ
.O

r
1 ∂ vφ vθ
C

vr
dφ φ = + cot φ +
©

r sin θ ∂ φ r r
 
1 1 ∂ vr ∂ vθ vθ 1 1 ∂ vr ∂ vφ vφ
drθ = + − drφ = + −
2 r ∂θ ∂r r 2 r sin θ ∂ φ ∂r r

1 1 ∂ vθ 1 ∂ vφ vφ
dθ φ = + − cot φ (2.121)
2 r sin θ ∂ φ r ∂θ r

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
94 C HAPTER 2. S TRAIN

rs
ee
s gin
t d le En

r
ba
ge ro or
Figure 2.28: Differential element in spherical coordinates.

eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 95

P ROBLEMS

Problem 2.1 – A deformation that takes place in a continuous medium has the
following consequences on the triangle shown in the figure below:
1. The segment OA increases its initial length in (1 + p).
2. The angle AOB decreases in q radians its initial value.

rs
3. The area increases its initial value in (1 + r).

ee
4. p, q, r, s 1.

s gin
The deformation is uniform and the z-axis is one of the principal directions of

t d le En
the deformation gradient tensor, which is symmetric. In addition, the stretch in
this direction is known to be λz = 1 + s. Obtain the infinitesimal strain tensor.

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O

Solution
C

A uniform deformation implies that the deformation gradient tensor (F) does
©

not depend on the spatial variables. Consequently, the strain tensor (E) and the
stretches (λ ) do not depend on them either. Also, note that the problem is to be
solved under infinitesimal strain theory.
The initial and final lengths of a segment parallel to the x-axis are related as
follows.
# A # A ⎫
OA f inal = λx dX = λx dX = λx OAinitial ⎬
O O =⇒ λx = 1 + p

OA f inal = (1 + p) OAinitial

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
96 C HAPTER 2. S TRAIN

Also, an initial right angle (the angle between the x- and y-axes) is related to its
corresponding final angle after the deformation through

π ⎪

initial angle = ⎬ q
2 =⇒ Δ Φ xy = −γxy = −2ε xy = −q =⇒ εxy = .
π ⎪
⎪ 2
final angle = + Δ Φ xy ⎭
2
In addition, F is symmetric and the z-axis is a principal direction, therefore
⎡ ⎤
∂ ux ∂ ux ∂ ux

rs
⎡ ⎤ ⎢1+ ∂x ∂y ∂z ⎥
⎢ ⎥

ee
F11 F12 0
not ⎢ ⎥ not not ⎢ ∂ uy ∂ uy ∂ uy ⎥
F ≡ ⎣ F12 F22 0 ⎦ ≡ 1 + J ≡ ⎢ ⎢ ⎥,

s gin
1+ ⎥
⎢ ∂ x ∂ y ∂ z ⎥
0 0 F33 ⎣ ∂ uz ∂ uz ∂ uz ⎦

t d le En
1+
∂x ∂y ∂z

r
ba
ge ro or
eS m
which reveals the nature of the components of the displacement vector,

ci
⎧  f

ra
C d P cs
⎪ ∂ ∂
bux (x, y) ,

a

⎪ u x u y
⎨ = = 0 =⇒
i
an an n

∂z ∂z uy (x, y) ,
y ha


⎪ ∂ u z ∂ uz

⎩ = = 0 =⇒ uz (z) .
le
liv or ec

∂x ∂y
M

.A

Then, the following components of the strain tensor can be computed.


m


d

1 ∂ ux ∂ u z
uu

εxz = + = 0 =⇒ εxz = 0
e

2 ∂z ∂x
X Th

er


tin

1 ∂ ux ∂ u z
εxz = + = 0 =⇒ εxz = 0
on

.O

2 ∂z ∂x ⎫
∂ uz ⎬
C

εzz = = λz − 1
∂z =⇒ εzz = s
©

λ = 1+s ⎭
z

In infinitesimal strain theory, F = 1 + ε + Ω , where Ω33 = 0 since the infinites-


imal rotation tensor is antisymmetric. Thus, Fzz = 1 + εzz results in Fzz = 1 + s .
Now, the relation between the initial and final areas is dA = |F| dA0 ·F−1 , where
the inverse tensor of F is calculated using the notation
⎡ ⎤
B11 B12 0 + ,
not ⎢ ⎥ C C
F ≡ ⎣ B12 B22 0 ⎦ with B−1 ≡
not 11 12
,
C12 C22
0 0 1+s

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 97

which yields the inverse tensor of F,


⎡ ⎤
C11 C12 0
⎢ ⎥
F−1 ≡ ⎢ 0 ⎥.
not
C C
⎣ 12 22 ⎦
1
0 0
1+s
The area differential vector is defined as
⎡ ⎤ ⎡ ⎤
0 0

rs
not ⎢ ⎥ not ⎢ ⎥
dA0 ≡ ⎣ 0 ⎦ =⇒ dA0 · F−1 ≡ ⎢ ⎣
0 ⎥.

ee
1
dA0 dA0

s gin
1+s

t d le En
Then, taking into account that |F| = Tr (εε ) + 1, and neglecting second-order
terms results in

r
ba

ge ro or
eS m
dA = (1 + r) dA0 ⎬

ci
f
=⇒ εyy = r − p .

ra
1
C d P cs
dA = (1 + p + s + εyy ) dA0 ⎭
b
a
1+s
i
an an n
y ha

Finally, since the strain tensor is symmetric,


le
⎡ ⎤
liv or ec

q
M

.A

⎢p 2
0⎥
not ⎢ q ⎥
ε ≡⎢ 0⎥⎥ .
m

⎢2 r− p
d

⎣ ⎦
uu
e

0 0 s
X Th

er
tin
on

.O
C

Problem 2.2 – A uniform deformation (F = F (t)) is produced on the tetrahe-


©

dron shown in the figure below, with the following consequences:

1. Points O, A and B do not move.


2. The volume of the solid becomes p times
its initial volume. √
3. The length of segment AC becomes p/ 2
times its initial length.
4. The final angle AOC has a value of 45◦ .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
98 C HAPTER 2. S TRAIN

Then,
a) Justify why the infinitesimal strain theory cannot be used here.
b) Determine the deformation gradient tensor, the possible values of p and the
displacement field in its material and spatial forms.
c) Draw the deformed solid.

Solution

rs
a) The angle AOC changes from 90◦ to 45◦ therefore, it is obvious that the

ee
deformation involved is not infinitesimal. In addition, under infinitesimal strain
theory Δ Φ 1 is satisfied and, in this problem, Δ Φ = π/4 ≈ 0.7854.

s gin
Observation: strains are dimensionless; in engineering, small strains are usually

t d le En
considered when these are of order 10−3 − 10−4 .

r
ba
ge ro or
eS m
b) The conditions in the statement of the problem must be imposed one by one:

ci
f
1. Considering that F (X,t) = F (t) and knowing that dx = F · dX, the latter

ra
C d P cs
b
can be integrated as

a
i
# # #
an an n
y ha

x= dx = FdX = F dX = F (t) · X + C (t)


le
liv or ec

⎡ ⎤ ⎡ ⎤
F11 F12 F13 C1
M

.A

not ⎢ ⎥ not ⎢ ⎥
with F ≡ ⎣ F21 F22 F23 ⎦ and C ≡ ⎣ C2 ⎦ ,
m

F31 F32 F33 C3


uu
e
X Th

er

which results in 12 unknowns. Imposing now the conditions in the statement,


tin

Point O does not move:


on

.O

⎡ ⎤ ⎡ ⎤ ⎡ ⎤
0 0 0
C

⎢ ⎥ ⎢ ⎥ not ⎢ ⎥
⎣ 0 ⎦ = [F] ⎣ 0 ⎦ + C =⇒ C ≡ ⎣ 0 ⎦
©

0 0 0

Point A does not move:


⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎧
a a a F11 ⎨ F11 = 1

⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣ 0 ⎦ = [F] ⎣ 0 ⎦ = ⎣ a F21 ⎦ =⇒ F21 = 0


0 0 a F31 F31 = 0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 99

Point B does not move:


⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎧
0 0 a F12 ⎨ F12 = 0

⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣ a ⎦ = [F] ⎣ a ⎦ = ⎣ a F22 ⎦ =⇒ F22 = 1


0 0 a F32 F32 = 0

Grouping all the information obtained results in


⎡ ⎤
1 0 F13
not ⎢ ⎥
F ≡ ⎣ 0 1 F23 ⎦ .

rs
ee
0 0 F33

s gin
2. The condition in the statement imposes that V f inal = pVinitial .

t d le En
Expression dV f = |F| dV0 allows to locally relate the differential volumes at
different instants of time. In this case, F is constant for each fixed t, thus, the

r
ba
ge ro or
expression can be integrated and the determinant of F can be moved outside the

eS m
ci
integral, # # #
f

ra
C d P cs
Vf = dV f = |F| dV0 = |F| dV0 = |F| V0 .
b
a
i
V V0 V0
an an n

Therefore, |F| = F33 = p must be imposed.


y ha

le
liv or ec

3. The condition in the statement imposes that lAC, f inal = √p lAC, initial .
2
M

.A

Since F is constant, the transformation is linear, that is, it transforms straight


lines into straight lines. Hence, AC in the deformed configuration must also be a
m

rectilinear segment. Then,


uu
e

⎡ ⎤⎡ ⎤ ⎡ ⎤
X Th

er
tin

1 0 F13 0 a F13
not ⎢ ⎥⎢ ⎥ ⎢ ⎥
xC = F · XC ≡ ⎣ 0 1 F23 ⎦ ⎣ 0 ⎦ = ⎣ a F23 ⎦ and
on

.O

0 0 F33 a ap
C

" " " "


lAC, f inal = lAC = " [a F13 , a F23 , ap] − [a, 0, 0] " = " [a (F13 − 1) , a F23 , ap] " =
! !
= (a (F13 − 1))2 + (a F23 )2 + (ap)2 = a (F13 − 1)2 + F23 2 + p2 =

p p √
= √ lAC = √ 2 a = p a .
2 2
Therefore,
!
(F13 − 1)2 + F23
2 + p2 = p ⇒ (F − 1)2 + F 2 = 0 ⇒ F = 1; F = 0
13 23 13 23

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
100 C HAPTER 2. S TRAIN

and the deformation gradient tensor results in


⎡ ⎤
1 0 1
not ⎢ ⎥
F ≡ ⎣0 1 0⎦ ,
0 0 p

such that only the value of p remains to be found.

4. The condition in the statement imposes that AOC f inal = 45◦ = π/4.

rs
Considering dX(1) ≡ [1, 0, 0] and dX(2) ≡ [0, 0, 1], the corresponding vectors
not not

ee
in the spatial configuration are computed as

s gin
⎡ ⎤⎡ ⎤ ⎡ ⎤
1 0 1 1 1
(1) not ⎢ ⎥⎢ ⎥ ⎢ ⎥

t d le En
(1)
dx = F · dX ≡ ⎣ 0 1 0 ⎦ ⎣ 0 ⎦ = ⎣ 0 ⎦ ,

r
0 0 p 0 0

ba
ge ro or
eS m
ci

⎤⎡ ⎤ ⎡ ⎤
f

ra
0 1 0 1
1
C d P cs
(2) not ⎢ ⎥⎢ ⎥ ⎢ ⎥
b
a
(2)
dx = F · dX ≡ ⎣ 0 1 0 ⎦ ⎣ 0 ⎦ = ⎣ 0 ⎦ .
i
an an n
y ha

0 0 p 1 p
le
liv or ec

Then, √
  dx(1) · dx(2) 2
M

.A


cos AOC f inal = cos 45 = "" (1) "" "" (2) "" =
dx dx 2
m

is imposed, with
uu
e

" " " "


" (2) " $
X Th

" (1) "


er
tin

"dx " = 1 , "dx " = 1 + p2 and dx(1) · dx(2) = 1


on

.O

such that √
C

1 2 1
$ = =√ =⇒ p = ±1 .
©

1 + p2 2 2
But |F| = p > 0, and, consequently, p = 1. Then, the deformation gradient tensor
is
⎡ ⎤
1 0 1
not ⎢ ⎥
F ≡ ⎣0 1 0⎦ .
0 0 1

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 101

The equation of motion is determined by means of x = F · X,


⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ ⎤
x 1 0 1 X X +Z
⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥
⎣y⎦ = ⎣0 1 0⎦⎣Y ⎦ = ⎣ Y ⎦ ,
z 0 0 1 Z Z

which allows determining the displacement field in material and spatial descrip-
tions as
⎡ ⎤ ⎡ ⎤

rs
Z z
not ⎢ ⎥ not ⎢ ⎥

ee
U (X,t) = x − X ≡ ⎣ 0 ⎦ and u (x,t) ≡ ⎣ 0 ⎦ .

s gin
0 0

t d le En
c) The graphical representation of the deformed tetrahedron is:

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O

Problem 2.3 – A uniform deformation is applied on the solid shown in the


figure below. Determine:
C

a) The general expression of the material description of the displacement field


U (X,t) in terms of the material displacement gradient tensor J.
b) The expression of U (X,t) when, in addition, the following boundary condi-
tions are satisfied:

UY = UZ = 0 , ∀ X, Y, Z
"
UX "X=0 = 0 , ∀ X, Y
"
UX "X=L = δ

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
102 C HAPTER 2. S TRAIN

c) The possible values (positive and negative) that δ may take. Justify the an-
swer obtained.
d) The material and spatial strain tensors and the infinitesimal strain tensor.
e) Plot the curves EXX − δ /L, exx − δ /L and εx − δ /L for all possible values
of δ , indicating every significant value.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
Solution
C d P cs
b
a
a) A uniform deformation implies that F (X,t) = F (t) , ∀t, X. The deformation
i
an an n

gradient tensor is related to the material displacement gradient tensor through


y ha

the expression F = 1 + J. Therefore, if F = F (t), then J = J (t). Taking into


le
account the definition of J and integrating its expression results in
liv or ec

# #
M

.A

∂ U (X,t)
J= =⇒ dU = J dX =⇒ dU = J dX
∂X
m

# #
uu
e

=⇒ dU = J dX =⇒ U = J · X + C (t) .
X Th

er
tin

where C (t) is an integration constant. Then, the general expression of the mate-
on

.O

rial description of the displacement field is


C

U (X,t) = J (t) · X + C (t) .

b) Using the previous result and applying the boundary conditions given in the
statement of the problem will yield the values of J and C.
Boundary conditions:
UY = UZ = 0 , ∀ X, Y, Z ⇒ Points only move in the X-direction.
"
UX "X=0 = 0 , ∀ Y, Z ⇒ The YZ plane at the origin is fixed.
"
UX "X=L = δ , ∀ Y, Z ⇒ This plane moves in a uniform manner
in the X-direction.
X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 103

If the result obtained in a) is written in component form, the equations and con-
clusions that can be reached will be understood better.

UX = J11 X + J12 Y + J13 Z +C1


UY = J21 X + J22 Y + J23 Z +C2
UZ = J31 X + J32 Y + J33 Z +C3

From the first boundary condition:

UY = 0 , ∀ X, Y, Z =⇒ J21 = J22 = J23 = C2 = 0

rs
UZ = 0 , ∀ X, Y, Z =⇒ J31 = J32 = J33 = C3 = 0

ee
s gin
From the second boundary condition:
"

t d le En
UX "X=0 = 0 , ∀ Y, Z =⇒ J12 = J13 = C1 = 0

r
ba
ge ro or
From the third boundary condition:

eS m
ci
" f

ra
δ
C d P cs
UX "X=L = δ , ∀ Y, Z =⇒ J11 L = δ ⇒ J11 =
b
a
L
i
an an n
y ha

Finally,
⎡ ⎤ ⎡ ⎤
⎡ ⎤ δ
le
δ
liv or ec

0 0 0 ⎢L ⎥X
⎢ L ⎥ not ⎢ ⎥ not ⎢ ⎥
M

.A

J≡⎢ ⎥
not
⎣0 0 0⎦; C ≡ ⎣0⎦ =⇒ U (X) = J · X + C ≡ ⎢ 0 ⎥ .
⎣ ⎦
m

0 0 0 0
d

0
uu
e
X Th

er
tin

c) In order to justify all the possible positive and negative values that δ may
on

.O

take, the condition |F| > 0 must be imposed. Therefore, the determinant of F
must be computed,
C

⎡ ⎤
δ
1+ 0 0
not ⎢ ⎥
F = 1+J ≡ ⎢
L ⎥ =⇒ |F| = 1 + δ > 0 =⇒ δ > −L .
⎣ 0 1 0 ⎦ L
0 0 1

d) To obtain the spatial and material strain tensors as well as the infinitesimal
strain tensor, their respective definitions must be taken into account.
1 
Spatial strain tensor: e= 1 − F−T · F−1
2

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
104 C HAPTER 2. S TRAIN

1 T 
Material strain tensor: E=
F ·F−1
2
1 T 
Infinitesimal strain tensor: ε = J ·J
2
Applying these definitions using the values of F and J calculated in b) and c),
the corresponding expressions are obtained.

⎡ ⎤
exx 0 0  -
⎢ ⎥ δ 1 δ2 δ 2

rs
not
e ≡ ⎣ 0 0 0 ⎦ with exx = + 1 +
L 2 L2 L

ee
0 0 0
⎡ ⎤ ⎡ ⎤

s gin
δ
EXX 0 0 ⎢L 0 0 ⎥
not ⎢ ⎥
⎥ with EXX = δ + 1 δ
2
not ⎢ ⎥

t d le En
E≡⎢ ⎣ 0 0 0 ⎦ ; ε ≡ ⎢ 0 0 0 ⎥
L 2L 2 ⎣ ⎦

r
ba
ge ro or
0 0 0 0 0 0

eS m
ci
f

ra
C d P cs
b
a
i
an an n

e) Plotting the curves EXX − δ /L , exx − δ /L and εx − δ /L together yields:


y ha

le
liv or ec

Here,
M

.A

• EXX is a second-order
parabola that contains the
m

origin and has its mini-


uu
e

mum at δ /L = −1, i.e., for


X Th

er

EXX = −1/2.
tin

• εx is the identity straight line


on

.O

(45◦ slope and contains the


C

origin).
©

• exx has two asymptotes, a


vertical one at δ /L = −1 and
a horizontal at exx = 1/2.

It can be concluded, then, that for small δ /L strains the three functions have a
very similar behavior and the same slope at the origin. That is, the same result
will be obtained with any of the definitions of strain tensor. However, outside
this domain (large or finite strains) the three curves are clearly different.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 105

E XERCISES

2.1 – Consider the velocity fields


   T
not x 2y 3z T X
not 2Y 3Z
v1 ≡ , , and v2 ≡ , , .
1+t 1+t 1+t 1+t 1+t 1+t
Determine:

rs
a) The material description of v1 and the spatial description of v2 (consider

ee
t = 0 is the reference configuration).

s gin
b) The density distribution in both cases (consider ρ0 is the initial density).
c) The material and spatial descriptions of the displacement field as well as

t d le En
the material (Green-Lagrange) and spatial (Almansi) strain tensors for the

r
velocity field v1 .

ba
ge ro or
eS m
d) Repeat c) for configurations close to the reference configuration (t → 0).

ci
f

ra
C d P cs
e) Prove that the two strain tensors coincide for the conditions stated in d).
b
a
i
an an n
y ha

2.2 – The equation of motion in a continuous medium is


le
liv or ec

x = X +Y t , y=Y , z=Z.
M

.A

Obtain the length at time t = 2 of the segment of material line that at time t = 1
m

is defined in parametric form as


d
uu
e

x (α) = 0 , y (α) = α 2 , z (α) = α 0≤α ≤1.


X Th

er
tin
on

.O

2.3 – Consider the material strain tensor


C

⎡ ⎤
©

0 tetX 0
⎢ ⎥
E≡⎢ 0 ⎥
not
⎦.
tX
⎣ te 0
0 0 tetY

Obtain the length at time t = 1 of the segment that at time t = 0 (reference


configuration) is straight and joins the points (1, 1, 1) and (2, 2, 2).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
106 C HAPTER 2. S TRAIN

2.4 – The equation of motion of a continuous medium is

x=X , y=Y , z = Z − Xt .
Calculate the angle formed at time t = 0 by the differential segments that at time
t = t are parallel to the x- and z-axes.

2.5 – The following information is known in relation to a certain displacement


field given in material description, U (X,Y, Z):
1) It is lineal in X, Y , Z.

rs
2) It is antisymmetric with re-

ee
spect to plane Y = 0, that is, the
following is satisfied:

s gin
U (X,Y, Z) = −U (X, −Y, Z)

t d le En
∀ X,Y, Z

r
ba
ge ro or
eS m
ci
3) Under said displacement field,
f

ra
C d P cs
the volume of the element in the
b
a
figure does not change, its an-
i
an an n

gle AOB remains constant,√ the


y ha

segment OB becomes 2 times


le
liv or ec

its initial length and the vertical


component of the displacement at
M

.A

point B is positive (wB > 0).


m

Determine:
d
uu
e

a) The most general expression of the given displacement field, such that condi-
X Th

er
tin

tions 1) and 2) are satisfied.


b) The expression of U when, in addition, condition 3) is satisfied. Obtain the
on

.O

deformation gradient tensor and the material strain tensor. Draw the de-
C

formed shape of the element in the figure, indicating the most significant
©

values.
c) The directions (defined by their unit vectors T) for which the deformation is
reduced to a stretch (there is no rotation).
NOTE: Finite strains must be considered (not infinitesimal ones).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 107

2.6 – The solid in the figure undergoes a uniform deformation such that points
A, B and C do not move. Assuming an infinitesimal strain framework,
a) Express the displacement field in terms of “generic” values of the stretches
and rotations.
b) Identify the null components of the strain tensor and express the rotation
vector in terms of the stretches.

In addition, the following is known:


1) Segment AE becomes (1 + p)
times its initial length.

rs
2) The volume becomes (1 + q)

ee
times its initial value.

s gin
3) The angle θ increases its value
in r (given in radians).

t d le En
Under these conditions, deter-

r
ba
mine:

ge ro or
eS m
ci
c) The strain tensor, the rotation
f

ra
C d P cs
vector and the displacement
b
a
i
field in terms of p, q and r.
an an n
y ha

NOTE: The values of p, q and


le
liv or ec

r are small and its second-order


infinitesimal terms can be ne-
M

.A

glected.
m

d
uu

2.7 – The solid in the figure undergoes a uniform deformation with the following
e

consequences:
X Th

er
tin

1) The x- and z-axes are both


on

material lines. Point A does


.O

not move.
C

2) The volume of the solid re-


©

mains constant.
3) The angle θxy remains con-
stant.
4) The angle θyz increases in r
radians.
5) The segment AF becomes
(1 + p) times its initial length.
6) The area of the triangle
ABE becomes (1 + q) its ini-
tial value.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
108 C HAPTER 2. S TRAIN

Then,
a) Express the displacement field in terms of “generic” values of the stretches
and rotations.
b) Identify the null components of the strain tensor and express the rotation
vector in terms of the stretches.
c) Determine the strain tensor, the rotation vector and the displacement field in
terms of p, q and r.
NOTE: The values of p, q and r are small and its second-order infinitesimal
terms can be neglected.

rs
ee
2.8 – The sphere in the figure undergoes a uniform deformation (F = const.)

s gin
such that points A, B and C move to positions A , B and C , respectively. Point
O does not move. Determine:

t d le En
a) The deformation gradient tensor in terms of p and q.

r
ba
ge ro or
b) The equation of the deformed external surface of the sphere. Indicate which

eS m
ci
type of surface it is and draw it.
f

ra
C d P cs
c) The material and spatial strain tensors. Obtain the value of p in terms of q
b
a
i
when the material is assumed to be incompressible.
an an n
y ha

d) Repeat c) using infinitesimal strain theory. Prove that when p and q are small,
the results of c) and d) coincide.
le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
CH.3. COMPATIBILITY
EQUATIONS
Multimedia Course on Continuum Mechanics
Overview

 Introduction Lecture 1

 Compatibility Conditions Lecture 2

 Compatibility Equations of a Potential Vector Field Lecture 3

 Compatibility Conditions for Infinitesimal Strains Lecture 4

 Integration of the Infinitesimal Strain Tensor Lecture 5

 Integration of the Deformation Rate Tensor Lecture 6

2
3.1 Compatibility Conditions
Ch.3. Compatibility Equations

3
Introduction
 Given a displacement field, the corresponding strain field is
found:
1  ∂U i ∂U j ∂U k ∂U k 
U ( X,t ) E=  + +  i, j ∈ {1, 2,3}
2 ∂X j ∂X i ∂X i ∂X j 

ij
 

1  ∂ui ∂u j 
u ( x,t ) ε ij = +  i, j ∈ {1, 2,3}
2  ∂x j ∂xi 
 

 Is the inverse possible?

ε ( x,t ) u ( x, t )

4
Compatibility Conditions
 Given an (arbitrary) symmetric second order tensor field, ε ( x,t ),
a displacement field, u ( x, t ), fulfilling ∇ s u(x, t ) = ε ( x, t ) cannot
always be obtained:
1  ∂ui ∂u j  6 PDEs OVERDETERMINED
ε ij = +  i, j ∈ {1, 2,3} 3 unknowns SYSTEM
2 ∂x j ∂xi 

 

 For ε ( x,t ) to match a symmetric strain tensor:


 It must be integrable.
 There must exist a displacement field from which it comes from.
REMARK
Given u ( x, t ) , there will always exist an associated
COMPATIBILITY CONDITIONS strain tensor, ε ( x, t ) , obtainable through
must be satisfied differentiation, which will automatically satisfy the
compatibility conditions.
5
Compatibility Conditions
 The compatibility conditions are the conditions a symmetric 2nd
order tensor must satisfy in order to be a strain tensor and, thus,
exist a displacement field which satisfies:
1  ∂ui ∂u j 
ε ij = +  i, j ∈ {1, 2,3}
2 ∂x j ∂xi 

 

 They guarantee the continuity of the continuous medium during the


deformation process.
E(X, t )

Incompatible
strain field

6
3.2 Compatibility Equations of a
Potential Vector Field
Ch.3. Compatibility Equations

7
Preliminary example: Potential Vector Field

 A vector field v ( x,t ) will be a potential vector field if there


exists a scalar function φ ( x,t ) (named potential function) such
that:
v ( x, t )= ∇φ ( x, t )
∂φ ( x, t )
vi ( x, t )
= i ∈ {1, 2,3}
∂xi

 Given a continuous scalar function φ ( x,t ) there will always exist a


potential vector field v ( x,t ) .
 Is the inverse true?
v ( x,t ) ∃ φ ( x,t ) such that ∇φ ( x, t ) =
v ( x, t )

8
Potential Field
v ( x,t ) φ ( x,t ) such that ∇φ ( x, t ) =
v ( x, t )

 In component form,
∂φ ( x, t ) ∂φ ( x, t )
vi ( x, t ) = vi ( x, t ) − =
0 i ∈ {1, 2,3} 1 3unknown
eqns.
∂xi ∂xi

OVERDETERMINED
 Differentiating once these expressions with respect to x j : SYSTEM

∂vi ∂ φ ( x, t )
2

= i, j ∈ {1, 2,3} 9 eqns.


∂x j ∂x j ∂xi

9
Schwartz Theorem
 The Schwartz Theorem about symmetry of second partial
derivatives guarantees that, given a continuous function
Φ ( x1 , x2 ,..., xn ) with continuous derivatives, the following holds
true:
∂Φ ∂Φ 2 2
= ∀i, j
∂xi ∂x j ∂x j ∂xi

10
Compatibility Equations
 Considering the Schwartz Theorem,
∂v x ∂ 2φ ∂v x ∂ 2φ ∂v x ∂ 2φ
= = =
∂x ∂x 2
∂y ∂x∂y ∂z ∂x∂z
∂v y ∂ 2φ ∂v y ∂ 2φ ∂v y ∂ 2φ
= = =
∂x ∂y∂x ∂y ∂y 2 ∂z ∂y∂z
∂v z ∂ 2φ ∂v z ∂ 2φ ∂v z ∂ 2φ
= = =
∂x ∂z∂x ∂y ∂z∂y ∂z ∂z 2
 In this system of 9 equations, only 6 different 2nd derivatives of the
unknown φ ( x,t ) appear: ∂ 2 φ ∂ 2 φ ∂ 2 φ ∂ 2 φ ∂ 2 φ ∂ 2φ
, , , , and
∂x 2
∂y 2
∂z 2
∂x∂y ∂x∂z ∂y∂z
 They can be eliminated and the following identities are obtained:
∂v x ∂v y ∂v x ∂v z ∂v y ∂v z
= = =
∂y ∂x ∂z ∂x ∂z ∂y
11
Compatibility Equations
 A scalar function φ ( x,t ) which satisfies ∇φ ( x, t ) =
v ( x, t ) will exist if
the vector field v ( x,t ) verifies:
∂v y ∂v x def
− ==
0 Sz
∂x ∂y eˆ1 eˆ 2 eˆ 3
Sx 
∂v x ∂v z def
  ∂ ∂ ∂
− ==
0 Sy where S ≡  S y  ≡ ≡ ∇×v
∂z ∂x  S  ∂x ∂y ∂z
∂v z ∂v y def  z v vy vz
− ==
0 Sx x
∂y ∂z
REMARK
A functional relation can be
INTEGRABILITY ∇ × v =0
(COMPATIBILITY)  established between these
EQUATIONS  ∂vi ∂v j three equations.
 ∂x − ∂x = 0 i, j ∈ {1, 2,3}
of a potential
vector field  j i ∇ ⋅ (∇ × v ) =
0

12
3.3 Compatibility Conditions for
Infinitesimal Strains
Ch.3. Compatibility Equations

13
Infinitesimal strains case
 The infinitesimal strain field can be written as:

 ∂u x 1  ∂u x ∂u y  1  ∂u x ∂u z  
  +   + 
 ∂x 2  ∂y ∂x  2  ∂z ∂x  
ε xx ε xy ε xz   
   ∂u y 
1 ∂u y ∂u z  
ε xy ε yy ε yz  =
ε  =  ×  +  
ε   ∂y 2  ∂z ∂y 
ε ε 
 xz yz zz 
 ∂u z 
 symmetrical × 
 ∂z 
 

6 PDEs
3 unknowns

14
Infinitesimal strains case
 The infinitesimal strain field can be written as:

∂u 1  ∂u x ∂u y 
ε xx − x 0
= ε xy −  = +  0
∂x 2  ∂y ∂x 
∂u 1  ∂u ∂u 
ε yy − y 0
= ε xz −  =
x
+ z 0
∂y 2  ∂z ∂x 
∂u 1  ∂u y ∂u z 
ε zz − z 0
= ε yz −  = +  0
∂z 2  ∂z ∂y 

6 PDEs The system will have a solution only if certain


3 unknowns compatibility conditions are satisfied.

15
Compatibility Conditions
 The compatibility conditions for the infinitesimal strain field are
obtained through double differentiation (single differentiation is
not enough).
 ∂u 
∂ 2  ε xx − x 
 ∂x 
= 6
∂ x , ∂y , ∂z , ∂xy, ∂xz , ∂yz
2 2 2
equations
  6x6=36
  equations
 1  ∂u y ∂u z  
∂  ε yz − 
2
+  
 
2  ∂z ∂y  
  =
6
∂ x 2 , ∂y 2 , ∂z 2 , ∂xy, ∂xz , ∂yz equations

16
Compatibility Conditions
 The compatibility conditions for the infinitesimal strain field are
obtained through: ∂ 2ε xx ∂ 3ux ∂ 2ε yz 1  ∂ 3u y ∂ 3u z 
= =  + 
∂x 2 ∂x 3 ∂x 2 2  ∂z∂x 2 ∂y∂x 2  18 equations for
ε xy ,ε xz ,ε yz
∂ ε xx
2
∂ ux3
∂ ε yz 1  ∂ u y ∂ u z 
2 3 3
= =  + 
∂y 2 ∂x∂y 2 ∂y 2 2  ∂z∂y 2 ∂y 3 

∂ 2ε xx ∂ 3u x ∂ 2ε yz 1  ∂ 3u y ∂ 3u z 
= ... =  + 
∂z 2 ∂x∂z 2 ∂z 2 2  ∂z 3 ∂y∂z 2 
18 equations for
ε xx ,ε yy ,ε zz ∂ 2ε xx ∂ 3u x ∂ 2ε yz 1  ∂ 3u y ∂ 3u z 
= =  + 
∂x∂y ∂x 2 ∂y ∂x∂y 2  ∂z∂x∂y ∂y 2 ∂x 

∂ 2ε xx ∂ 3u x ∂ 2ε yz 1  ∂ 3u y ∂ 3u z 
= =  + 
∂x∂z ∂x 2 ∂z ∂x∂z 2  ∂z 2 ∂x ∂y∂x∂z 

∂ 2ε xx ∂ 3u x ∂ 2ε yz 1  ∂ 3u y ∂ 3u z 
= =  + 
∂y∂z ∂x∂y∂z ∂y∂z 2  ∂z 2∂y ∂y 2∂z 

17
Compatibility Conditions
 All the third derivatives of u x , u y and u z appear in the
equations: 3
∂ ux
= 10 derivatives
∂x , ∂x y, ∂x z , ∂y , ∂y x, ∂y z , ∂z , ∂z x, ∂z y, ∂xyz
3 2 2 3 2 2 3 2 2

∂ 3u y
= 10 derivatives
∂x , ∂x y, ∂x z , ∂y , ∂y x, ∂y z , ∂z , ∂z x, ∂z y, ∂xyz
3 2 2 3 2 2 3 2 2

∂ 3u z
= 10 derivatives
∂x , ∂x y, ∂x z , ∂y , ∂y x, ∂y z , ∂z , ∂z x, ∂z y, ∂xyz
3 2 2 3 2 2 3 2 2

which constitute 30 of the unknowns in the system of 36 equations:


 ∂ 3u ∂ 2
ε ij 
fn  i
, =  0 n ∈ {1, 2,...,36}
 ∂x j ∂xk ∂xl ∂xk ∂xl 
 

18
30
Compatibility Equations
∂ 3ui
 Eliminating the 30 unknowns , , 6 equations (involving
∂x j ∂xk ∂xl
only strain derivatives) are obtained:
 def ∂ 2ε yy ∂ 2ε ∂ 2
ε yz COMPATIBILITY
 S xx = + zz
−2 = 0 EQUATIONS
 ∂z 2
∂y 2
∂y∂z for the infinitesimal
 def
∂ ε zz ∂ ε xx
2 2
∂ 2ε xz strain tensor
 S yy = ∂x 2 + ∂z 2 − 2 ∂x∂z = 0
 def ∂ 2ε ∂ ε ∂ ε xy S = ∇ × (ε × ∇ ) = 0
2 2
 S zz = xx
+
yy
−2 =0
 ∂ y 2
∂ x 2
∂ x ∂ y
 def ∂ 2ε zz ∂  ∂ε yz ∂ε xz ∂ε xy 
 S xy = − + 
 ∂x
+ −  = 0
 ∂x ∂y ∂ z  ∂ y ∂z 
 def ∂ ε yy ∂  ∂ε yz ∂ε xz ∂ε xy 
2

 S xz = − + 

− +  = 0
 ∂ ∂
x z ∂  ∂
y x ∂ y ∂ z 
 def ∂ 2ε xx ∂  ∂ε yz ∂ε xz ∂ε xy 
 S yz = − + −

+ +  = 0
 ∂y∂z ∂x  ∂x ∂y ∂z 
19
Compatibility Equations
 The six equations are not functionally independent. They satisfy
the equation,
∇ ⋅ S = ∇ ⋅ (∇ × (ε × ∇ )) = 0

 In indicial notation:

∂S xx ∂S xy ∂S xz
+ + =
0
∂x ∂y ∂z
∂S xy ∂S yy ∂S yz
+ + =
0
∂x ∂y ∂z
∂S xz ∂S yz ∂S zz
+ + =
0
∂x ∂y ∂z

20
Compatibility Equations
 The compatibility equations can be expressed in terms of the
permutation operator, eijk .

= ε ij ,qr 0 m, l ∈ 1, 2,3


S ml emjqelir=

 Or, alternatively:

ε ij ,kl + ε kl ,ij − ε ik , jl −=
ε jl ,ik 0 i, j , k , l ∈ {1, 2,3}

REMARK
Any linear strain tensor (1st order polynomial) with respect to
the spatial variables will be compatible and, thus, integrable.

21
3.4 Integration of the Infinitesimal
Strain Tensor
Ch.3. Compatibility Equations

22
Preliminary Equations
 Rotation tensor Ω ( x,t ) :
 1
=Ω skew(u ⊗ = ∇) (u ⊗ ∇ − ∇ ⊗ u)
 2

= 1  ∂ui ∂u j 
Ω  −  i, j ∈ {1, 2,3}
 ij 2  ∂x j ∂xi 

 Rotation vector θ ( x,t ) :

θ1   −Ω 23   −Ω yz   0 −θ3 θ2 
 
θ = ∇ × u = θ 2  =  −Ω  =
[Ω(θ )] =  θ3 −θ1 
1
 31  −Ω
 zx  0
2
θ3   −Ω12     −θ 2 θ1 0 
 −Ω xy 

23
Preliminary Equations
 Differentiating Ω ( x,t ) with respect to xk :

1  ∂ui ∂u j  ∂Ωij 1 ∂  ∂ui ∂u j 


=
Ωij  −  =  − 
2  ∂x j ∂xi  ∂xk 2 ∂xk ∂
 j
x ∂xi 

 Adding and subtracting the term 1 ∂ 2


uk :
2 ∂xi ∂x j
∂Ωij 1 ∂  ∂ui ∂u j  1 ∂ 2uk 1 ∂ 2uk
=  − + − =
∂xk 2 ∂xk  ∂x j ∂xi  2 ∂xi ∂x j 2 ∂xi ∂x j
∂ 1  ∂ui ∂uk  ∂ 1  ∂u j ∂uk  ∂ε ik ∂ε jk
=  + −  +  = −
∂x j 2  ∂xk ∂xi  ∂xi 2  ∂xk ∂x j  ∂x j ∂xi
= ε ik = ε jk
24
Preliminary Equations
 Using the previous results, the derivative of θ ( x,t ) is obtained:
 ∂θ 2 ∂Ω zx ∂ε xx ∂ε xz
 ∂θ1 ∂Ω yz ∂ε xz ∂ε xy =
− =−
=
− =−  ∂x ∂x ∂z ∂x
 
 ∂x ∂x ∂ y ∂ z
 ∂θ 2 ∂Ω zx ∂ε xy ∂ε yz

 ∂θ1 ∂Ω yz ∂ε yz ∂ε yy ∇θ 2  =
− =−
∇θ1  =
− =−  ∂y ∂y ∂z ∂x
 ∂y ∂y ∂ y ∂ z
 ∂θ 2 ∂Ω zx ∂ε xz ∂ε zz
 ∂θ ∂Ω ∂ε zz ∂ ε  =
− =−
 1
=
− yz
=− zy
 ∂z ∂z ∂z ∂x

 ∂ z ∂ z ∂y ∂ z
 ∂θ3 ∂Ω xy ∂ε xy ∂ε xx
 = − =−
 ∂x ∂x ∂x ∂y
 ∂θ3 ∂Ω xy ∂ε yy ∂ε xy
∇θ 3  = − =−
 ∂y ∂y ∂x ∂y
 ∂θ ∂Ω xy ∂ε yz ∂ε xz
 3
= − =−
 ∂z ∂z ∂x ∂y
25
Preliminary Equations
 Considering the displacement gradient tensor J ( x,t ) ,
∂u ( x, t )
J= = ε+Ω
∂x = ε ij
∂ui 1  ∂ui ∂u j  1  ∂ui ∂u j 
J= =  + +  − =  ε ij + Ωij i, j ∈ {1, 2,3}
∂x j 2  ∂x j ∂xi  2  ∂x j ∂xi 
ij

= Ωij
 Introducing the definition of θ ( x,t ) , the components of J ( x,t )
are rewritten:
= j 1= j 2= j 3
∂u x ∂u x ∂u x
i = 1: =ε xx =ε xy − θ3 =ε xz + θ 2
∂x ∂y ∂z
∂u y ∂u y ∂u y
i = 2: =ε xy + θ3 =ε yy =ε yz − θ1
∂x ∂y ∂z
∂u z ∂u z ∂u z
i = 3: =ε xz − θ 2 =ε yz + θ1 =ε zz
∂x ∂y ∂z
26
Integration of the Strain Field
 The integration of the strain field ε ( x,t ) is performed in two
steps:
1. Integration of derivative of θ ( x,t ) using the1st order PDE system
derived for ∇θ1 , ∇θ 2 and ∇θ3 . The solution will be of the type:
θ=i θi ( x, y, z, t ) + ci ( t ) i ∈ {1, 2,3}
The integration constants ci ( t ) can be obtained knowing the value of
the rotation vector in some points of the medium (boundary conditions).

2. Known ε ( x,t ) and θ ( x,t ) , u is integrated using the 1st order PDE
REMARK system derived for u ⊗ ∇ . The solution will be:
If the compati- ui = ui ( x, y, z , t ) + ci′ ( t ) i ∈ {1, 2,3}
bility equations
The integration constants ci′ ( t ) can be obtained knowing the value of
are satisfied, the displacements in some point of space (boundary conditions)
these equations
will be integra-
ble. 27
Integration of the Strain Field
 The integration constants that appear imply that an integrable
strain tensor ε ( x,t ) will determine the movement in any instant not
of
not
time except for a rotation c(t ) = θˆ (t ) and a translation c′(t ) = uˆ (t ) :

 ( x, t ) θ ( x, t ) + θ ( t )
θ=  ˆ
ε ( x, t ) 
( x, t ) u ( x, t ) + uˆ ( t )
u=
 A displacement field can be constructed from this uniform rotation
and translation: u∗ (= ˆ (θˆ(t )) ⋅ x + uˆ (t )
x, t ) Ω
u∗ ⊗ ∇ = Ω
ˆ
 S * ∗ ∗
 ∇ ( u =
) 1
2 (u ⊗ ∇ + (u ⊗ ∇ )T
)= 1
2 (Ω
ˆ +Ω
ˆ T )= 0

 This corresponds to a rigid solid movement.

28
3.5 Integration of the Deformation
Rate Tensor
Ch.3. Compatibility Equations

29
Compatibility Equations in a
Deformation Rate Field
 There is a correspondence between
u v
ε (u) d( v )
1  ∂ui ∂u j  1  ∂vi ∂v j 
=ε ij  +  =dij  + 
2  ∂x j ∂xi  2  ∂x j ∂xi 

1  ∂ui ∂u j  1  ∂vi ∂v j 
=
Ωij  −  =
w ij  − 
2  ∂x j ∂xi  2  ∂x j ∂xi 
1 1
=
θ ∇ ×u =
ω ∇×v
2 2

 The concept of compatibility conditions can be extended to


deformation rate tensor d ( v ) .

30
Example
Obtain the velocity field corresponding to the deformation rate tensor:
 0 tety 0 
 
d ( x, t )  = tety 0 0 
 0 0 tetz 
 

such that In point (1, 1, 1) the following conditions is fulfilled:


 2e t   0 
   
v ( x, t )
1
=  et  ω ( x, t ) = ∇ × v=  0 
x=(1,1,1)  t  x=(1,1,1) 2 −tet 
e   

31
 0 tety 0 
 
Example - Solution d ( x, t )  = tety
 0

0
0
0
tetz


u
Consider the correspondence: v
ε(u) d( v )
1 1
=
θ ∇ ×u =
ω ∇×v
2 2
Take the expressions derived for ∇θ1 , ∇θ 2 and ∇θ3 substitute θ ( x,t ) with ω ( x,t )
and ε ( x,t ) with d ( x,t ) :
 ∂ω1 ∂d xz ∂d xy
 = − =0−0
 ∂x ∂y ∂z

 ∂ω1 ∂d yz ∂d yy
∇ω1  = − = 0−0 ω1 ( t ) =
C1 ( t )
 ∂y ∂y ∂z
 ∂ω ∂d ∂d zy
 1 =zz − =0−0
 ∂z
 ∂y ∂z

32
 0 tety 0 
 
Example - Solution d ( x, t )  = tety
 0

0
0
0
tetz


 ∂ω2 ∂d xx ∂d xz
 ∂x = ∂z

∂x
=0−0

 ∂ω2 ∂d xy ∂d yz
∇ω2  = − = 0−0 ω2 ( t ) =
C2 ( t )
 ∂y ∂z ∂x
 ∂ω2 ∂d xz ∂d zz
 = − =0−0
 ∂z ∂z ∂x
 ∂ω3 ∂d xy ∂d xx
 = − =0−0
 ∂x ∂x ∂y
 ∂ω3 ∂d yy ∂d xy
∇ω3  = − = 0 − t 2 ety ω3 ( y, t ) =
∫ −t 2 ty
e dy =
−te ty
+ C3 ( t )
 ∂y ∂x ∂y
 ∂ω ∂d yz ∂d
 3 = − xz = 0−0
 ∂z ∂x ∂y

33
ω1 =C1 ( t )

Example - Solution ω2 =C2 ( t )


ω3 =−tety + C3 ( t )

For point (1, 1, 1) :


 0 
{ ( )} 2 ∇ × v=  0 
ω x, t =
1
−tet 
 
So,
ω1 = 0 = C1 ( t )
C1 ( t ) = 0
ω2 = 0 = C2 ( t ) C2 ( t ) = 0
 
ω3 =−tet = −tety + C3 ( t )  C3 ( t ) = 0
  x=(1,1,1)
Therefore, for any point,
 0 
{ω ( x, t )} =  0 
−tety 
 
34
 0   0 tety 0 
 
{ω ( x, t )} =  0  ; d ( x, t ) = tety 
Example - Solution −tety 
   0

0
0
0 
tetz 

Taking the expressions = j 1= j 2= j 3


∂v x ∂v x ∂v x
i = 1: = d xx = d xy − ω3 = d xz + ω2
∂x ∂y ∂z
∂v y ∂v y ∂v y
i = 2: = d xy + ω3 = d yy = d yz − ω1
∂x ∂y ∂z
∂v z ∂v z ∂v z
i = 3: = d xz − ω2 = d yz + ω1 = d zz
∂x ∂y ∂z
The components of the velocities can be obtained:
∂v x
= d= 0
∂x
xx

∂v x
= d xy − ω= tety − ( −tety=
) 2tety v x ( y=
,t) ∫ 2te =
ty
dy 2ety + C1′ ( t )
∂y
3

∂v x
= d xz + ω2 = 0 + 0
∂z

35
 0   0 tety 0 
 
{ω ( x, t )} =  0  ; d ( x, t ) = tety 
Example - Solution −tety 
   0

0
0
0 
tetz 

The components of the velocities can be obtained:


∂v y
= d xy + ω=
∂x
3 te ty
+ ( −te =
ty
) 0

∂v y
= d= 0 v y ( t ) = C2′ ( t )
∂y
yy

∂v y
= d yz − ω1 = 0 − 0
∂z

∂v z
= d xz − ω2 = 0 − 0
∂x
∂v z
= d yz + ω1 = 0 + 0 v z ( z , t=
) ∫ te tz
=
dz e tz
+ C3′ ( t )
∂y
∂v z
= d= tetz
∂z
zz

36
v x 2ety + C1′ ( t )
=

Example - Solution v y = C2′ ( t )


v=
z etz + C3′ ( t )

For point (1, 1, 1) :  2e t 


 t 
{ v ( x, t )} e 
=
 et 
 

So,  
= et  2ety + C1′ ( t ) 
v x 2=
  x=(1,1,1) C1′ ( t ) = 0
v=y e=
t
C2′ ( t ) C2′ ( t ) = et
 tz  C3′ ( t ) = 0
v=
z e=t
 e + C3
′ ( t ) 
  x=(1,1,1)
Therefore, for any point, 2ety 
 t 
{ ( )}  e 
v x , t =
 etz 
 
37
Chapter 3
Compatibility Equations

rs
ee
s gin
3.1 Introduction

t d le En
Given a sufficiently regular displacement field U (X,t), it is always possible to

r
ba
ge ro or
eS m
find the corresponding strain field (for example, the Green-Lagrange strain field)

ci
f
by differentiating this strain field with respect to its coordinates (in this case, the

ra
C d P cs
material ones)1 ,
b
a
 
i
an an n

1 ∂Ui ∂U j ∂Uk ∂Uk not 1  


Ei j = + + = Ui, j +U j,i +Uk,i Uk, j
y ha

2 ∂ X j ∂ Xi ∂ Xi ∂ X j 2 (3.1)
le
i, j ∈ {1, 2, 3}.
liv or ec
M

.A

In the infinitesimal strain case, given a displacement field u (x,t), the strain
field
m

 
d

1 ∂ ui ∂ u j not 1
uu

εi j = + = (ui, j + u j,i ) i, j ∈ {1, 2, 3}


e

(3.2)
2 ∂ x j ∂ xi
X Th

2
er
tin

is obtained.
on

.O

The question can be formulated in reverse, that is, given a strain field ε (x,t),
is it possible to find a displacement field u (x,t) such that ε (x,t) is its infinites-
C

imal strain tensor? This is not always possible and the answer provides the so-
called compatibility equations.
Expression (3.2) constitutes a system of 6 (due to symmetry) partial differen-
tial equations (PDEs) with 3 unknowns: u1 (x,t), u2 (x,t), u3 (x,t). This system
is overdetermined because there exist more conditions than unknowns, and it
may not have a solution.
Therefore, for a second-order symmetric tensor ε (x,t) to correspond to a
strain tensor (and, thus, be integrable and there exist a displacement field from
which it comes) it is necessary that this tensor verifies certain conditions. These
conditions are denominated compatibility conditions or equations and guarantee
not
1 Here, the simplified notation ∂Ui /∂ X j = Ui, j is used.

109
110 C HAPTER 3. C OMPATIBILITY E QUATIONS

Figure 3.1: Non-compatible strain field.

rs
the continuity of the continuous medium during the deformation process (see

ee
Figure 3.1).

s gin
t d le En
Definition 3.1. The compatibility conditions are conditions that a
second-order tensor must satisfy in order to be a strain tensor and,

r
ba
ge ro or
therefore, for there to exist a displacement field from which it comes.

eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

Remark 3.1. Note that, to define a strain tensor, the 6 components of


le
a symmetric tensor cannot be written arbitrarily. These must satisfy
liv or ec

the compatibility conditions.


M

.A
m

d
uu
e
X Th

Remark 3.2. Given a displacement field, one can always obtain,


er
tin

through differentiation, an associated strain field that automatically


satisfies the compatibility conditions. Therefore, in this case, there is
on

.O

no sense in verifying that the compatibility conditions are satisfied.


C

3.2 Preliminary Example: Compatibility Equations of a


Potential Vector Field
A given vector field v (x,t) is a potential field if there exists a scalar function
φ (x,t) (named potential function) such that its gradient is v (x,t),


⎨ v (x,t) = ∇φ (x,t) ,
∂ φ (x,t) (3.3)

⎩ vi (x,t) = i ∈ {1, 2, 3} .
∂ xi

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Preliminary Example: Compatibility Equations of a Potential Vector Field 111

Therefore, given a scalar (continuous) function φ (x,t), it is always possible to


define a potential vector field v (x,t) such that the scalar function is its potential,
as defined in (3.3).
Now, the reverse question is posed: given a vector field v (x,t), does there
exist a scalar function φ (x,t) such that ∇φ (x,t) = v (x,t)? This is written in
component form as
∂φ ∂φ
vx = =⇒ vx − =0,
∂x ∂x
∂φ ∂φ
vy = =⇒ vy − =0, (3.4)
∂y ∂y

rs
∂φ ∂φ
vz = =⇒ vz − =0,

ee
∂z ∂z

s gin
which corresponds to a system of PDEs with 3 equations and 1 unknown
(φ (x,t)), thus, the system is overdetermined and may not have a solution.

t d le En
Differentiating once (3.4) with respect to (x, y, z) yields

r
ba
∂ vx ∂ 2 φ ∂ vx ∂ 2φ ∂ vx ∂ 2φ

ge ro or
eS m
= 2 , = , = ,

ci
∂x ∂x ∂y ∂ x∂ y ∂z ∂ x∂ z
f

ra
C d P cs
∂ vy ∂ 2φ ∂ vy ∂ 2 φ ∂ vy ∂ 2φ
b
a
= , = 2 , = , (3.5)
i
∂x ∂ y∂ x ∂y ∂y ∂z ∂ y∂ z
an an n
y ha

∂ vz ∂ 2φ ∂ vz ∂ 2φ ∂ vz ∂ 2 φ
= , = , = 2 ,
le
liv or ec

∂x ∂ z∂ x ∂y ∂ z∂ y ∂z ∂z
M

.A

which represents a system of 9 equations. Considering the equality of mixed


partial derivatives, it is observed that 6 different functions (second derivatives)
m

of the unknown φ are involved in these 9 equations,


uu
e
X Th

∂ 2φ ∂ 2φ ∂ 2φ ∂ 2φ ∂ 2ϕ ∂ 2φ
er
tin

, , , , and . (3.6)
∂ x2 ∂ y2 ∂ z2 ∂ x∂ y ∂ x∂ z ∂ y∂ z
on

.O

So, they can be removed from the original system (3.5) and 3 relations, named
C

compatibility conditions, can be established between the first partial derivatives


©

of the components of v (x,t).


Hence, for there to exist a scalar function φ (x,t) such that ∇φ (x,t) = v (x,t),
the given vector field v (x,t) must satisfy the following compatibility conditions.
⎫  
∂ v y ∂ vx ⎪ ⎡ ⎤  ê ê ê 
= 0 = Sz ⎪
de f
− ⎪

∂x ∂y ⎪
⎪ Sx  1 2 3
⎬ ⎢ ⎥  
∂ vx ∂ vz de f not
⎢ ⎥  ∂ ∂ ∂  not not
− = 0 = Sy ⎪ where S ≡ ⎣ Sy ⎦ ≡   ≡ rot v = ∇ × v

∂z ∂x ⎪

⎪  ∂x ∂y ∂z 
∂ vz ∂ vy de f ⎪
⎪ Sz  
− = 0 = Sx ⎭  vx vy vz 
∂y ∂z
(3.7)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
112 C HAPTER 3. C OMPATIBILITY E QUATIONS

In consequence, from (3.7), the compatibility equations can be written as



⎨∇×v = 0
Compatibility equations
of a potential vector field ⎩ ∂ vi − ∂ v j = 0
(3.8)
i, j ∈ {1, 2, 3}
∂ x j ∂ xi

Remark 3.3. The 3 compatibility equations (3.7) or (3.8) are not in-
dependent of one another and a functional relation can be established

rs
between them. Indeed, applying the condition that the divergence of

ee
the rotational of a vector field is null2 , ∇ · (∇ × v) = 0 .

s gin
t d le En

r
3.3 Compatibility Conditions for Infinitesimal Strains

ba
ge ro or
eS m
ci
f
Consider the infinitesimal strain field ε (x,t) with components

ra
C d P cs
 
b
a
1 ∂ ui ∂ u j not 1
i
an an n

εi j = + = (ui, j + u j,i ) i, j ∈ {1, 2, 3} , (3.9)


2 ∂ x j ∂ xi
y ha

2
le
liv or ec

which may be written in matrix form as


⎡    ⎤
M

.A

∂ u x 1 ∂ u x ∂ uy 1 ∂ ux ∂ uz
⎡ ⎤ ⎢ ∂x 2 ∂y + ∂x + ⎥
m

εxx εxy εxz ⎢ 2 ∂z ∂x ⎥


d

⎢  ⎥
⎢ ⎥ ⎢ 1 ∂ uy ∂ u z ⎥
uu

∂ uy
e

[εε ] = ⎣ εxy εyy εyz ⎦ = ⎢ × + ⎥.


⎢ ∂y ⎥
X Th

∂y 2 ∂z
er
tin

εxz εyz εzz ⎢ ⎥


⎣ ∂ uz ⎦
(symm) ×
on

.O

∂z
C

(3.10)
©

Due to the symmetry in (3.10), only 6 different equations are obtained,


 
∂ ux 1 ∂ ux ∂ uy
εxx − =0, εxy − + =0,
∂x 2 ∂y ∂x
 
∂ uy 1 ∂ ux ∂ uz
εyy − =0, εxz − + =0, (3.11)
∂y 2 ∂z ∂x
 
∂ uz 1 ∂ uy ∂ u z
εzz − =0, εyz − + =0.
∂z 2 ∂z ∂y
2 A theorem of differential geometry states that the divergence of the rotational of any field
is null, ∇ · [∇ × (•)] = 0.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Compatibility Conditions for Infinitesimal Strains 113

Equation (3.11) is a system of 6 PDEs with 3 unknowns, which are the compo-
nents of the displacement vector u (x,t) ≡ [ux , uy , uz ]T . In general, this problem
not

will not have a solution unless certain compatibility conditions are satisfied. To
obtain these conditions, the equations in (3.11) are differentiated twice with re-
spect to their spatial coordinates,
 
∂ ux
∂ εxx −
2
∂x
= 6 equations
∂ x , ∂ y , ∂ z , ∂ xy, ∂ xz, ∂ yz
2 2 2
.. .. (3.12)
 .   .

rs
1 ∂ uy ∂ u z
∂ 2 εyz − +

ee
2 ∂z ∂y
= 6 equations ,

s gin
∂ x2 , ∂ y2 , ∂ z2 , ∂ xy, ∂ xz, ∂ yz

t d le En
providing a total of 36 equations,
 

r
ba
∂ 2 εxx ∂ 3 ux ∂ 2 εyz 1 ∂ 3 uy ∂ 3 uz

ge ro or
eS m
= = +

ci
∂ x2 ∂ x3 ∂ x2 2 ∂ z∂ x2 ∂ y∂ x2
f  

ra
C d P cs
∂ εxx
2 ∂ 3 ux ∂ 2 εyz 1 ∂ 3 uy ∂ 3 uz
b
a
= = +
i
∂ y2 ∂ x∂ y2 ∂ y2 2 ∂ z∂ y2 ∂ y3
an an n

 
y ha

∂ εxx
2 ∂ 3 ux ∂ εyz 1 ∂ uy
2 3
∂ 3 uz
= ··· = +
le
∂ z2 ∂ x∂ z2 ∂ z2 2 ∂ z3 ∂ y∂ z2
liv or ec

 
∂ 2 εxx ∂ 3 ux ∂ 2 εyz 1 ∂ 3 uy ∂ 3 uz
M

.A

= 2 = + 2 (3.13)
∂ x∂ y ∂ x ∂ y ∂ x∂ y 2 ∂ z∂ x∂ y ∂ y ∂ x
 
m

∂ 2 εxx ∂ 3 ux ∂ 2 εyz 1 ∂ 3 uy ∂ 3 uz
d

= 2 = +
uu

∂ x∂ z ∂ x ∂ z ∂ x∂ z 2 ∂ z2 ∂ x ∂ y∂ x∂ z
e

 
X Th

er

∂ 2 εxx ∂ 3 ux ∂ 2 εyz 1 ∂ 3 uy ∂ 3 uz
tin

= = +
∂ y∂ z ∂ x∂ y∂ z ∂ y∂ z 2 ∂ z2 ∂ y ∂ y2 ∂ z
  
on

  
.O

(18 eqns for εxx , εyy , εzz ) (18 eqns for εxy , εxz , εyz )
C

All the possible third derivatives of each component of the displacements ux , uy


and uz are involved in these 36 equations. Thus, there are 30 different derivatives,
∂ 3 ux
= 10 derivatives ,
∂ x3 , ∂ x2 y, ∂ x2 z, ∂ y3 , ∂ y2 x, ∂ y2 z, ∂ z3 , ∂ z2 x, ∂ z2 y, ∂ xyz
∂ 3 uy
= 10 derivatives ,
∂ x3 , ∂ x2 y, ∂ x2 z, ∂ y3 , ∂ y2 x, ∂ y2 z, ∂ z3 , ∂ z2 x, ∂ z2 y, ∂ xyz
∂ 3 uz
= 10 derivatives ,
∂ x3 , ∂ x2 y, ∂ x2 z, ∂ y3 , ∂ y2 x, ∂ y2 z, ∂ z3 , ∂ z2 x, ∂ z2 y, ∂ xyz
(3.14)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
114 C HAPTER 3. C OMPATIBILITY E QUATIONS

which constitute the 30 unknowns in the system of 36 equations


 
∂ 3 ui ∂ 2 εi j
fn , n ∈ {1, 2 ... 36} (3.15)
∂ x j ∂ xk ∂ xl ∂ xk ∂ xl
  
30
defined in (3.13). Therefore, the 30 unknowns, which are the displacement
derivatives ∂ 3 ui /(∂ x j ∂ xk ∂ xl ), can be eliminated from this system and 6 equa-
tions are obtained. In these equations, the third derivatives mentioned above
do not appear, but there will be 21 second derivatives of the strain tensor

rs
∂ 2 εi j /(∂ xk ∂ xl ). After the corresponding algebraic operations, the resulting equa-

ee
tions are

s gin


⎪ de f ∂ εyy
2
∂ 2 εzz ∂ 2 εyz

t d le En

⎪ = + − =0

⎪ S xx 2

⎪ ∂ z2 ∂ y2 ∂ y∂ z

r

ba
ge ro or

eS m

⎪ de f ∂ εzz
2 ∂ 2 εxx ∂ 2 εxz

ci
⎪ S = + − 2 =0
⎪ yy
f

ra

⎪ ∂ x2 ∂ z2 ∂ x∂ z
C d P cs


b
a

⎪ ∂ 2 εyy ∂ 2 εxy
de f ∂ εxx
2

i

an an n

⎨ Szz = + − 2 =0
∂ y2 ∂ x2 ∂ x∂ y
y ha

Compatibility
equations ⎪   (3.16)
le

⎪ ∂ 2 εzz ∂ ∂ εyz ∂ εxz ∂ εxy
liv or ec


⎪ S
de f
= − + + − =0

⎪ xy
∂ x∂ y ∂ z ∂ x ∂y ∂z
M


.A


⎪  



⎪ de f ∂ εyy
2
∂ ∂ εyz ∂ εxz ∂ εxy
m


⎪ Sxz = − + − + =0
d


⎪ ∂ x∂ z ∂ y ∂ x ∂y ∂z
uu



e


⎪  
X Th

⎪ de f ∂ εxx
2 ∂ ∂ εyz ∂ εxz ∂ εxy
er


tin

⎩ Syz = − + − + + =0
∂ y∂ z ∂ x ∂x ∂y ∂z
on

.O

which constitute the compatibility equations for the infinitesimal strain tensor ε .
C

The compact expression corresponding to the 6 equations in (3.16) is



Compatibility equations
for the infinitesimal S = ∇ × (εε × ∇) = 0 (3.17)
strain tensor

Another way of expressing the compatibility conditions


 (3.16)
 is in terms of
the three-index operator named permutation operator ei jk . In this case, the
compatibility equations can be written as

Smn = em jq enir εi j,qr = 0 . (3.18)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Compatibility Conditions for Infinitesimal Strains 115

Remark 3.4. The 6 equations (3.16) are not functionally independent


and, taking again into account the fact that the divergence of the rota-
tional of a field is intrinsically null, the following functional relations
can be established between them.


⎪ ∂ Sxx ∂ Sxy ∂ Sxz

⎪ + + =0

⎪ ∂x ∂y ∂z



∇ · S = ∇ · (∇ × (εε × ∇)) = 0 =⇒ ∂ Sxy ∂ Syy ∂ Syz
+ + =0

rs

⎪ ∂x ∂y ∂z

ee



⎪ ∂S ∂S ∂S
⎩ xz + yz + zz = 0

s gin
∂x ∂y ∂z

t d le En

r
ba
ge ro or
eS m
ci
f

ra
Remark 3.5. The three-index operator denominated permutation op-
C d P cs
b
a
erator is given by
i
an an n


y ha


⎪ 0 → if an index is repeated,

⎪ i = j or i = k or j = k
le


liv or ec


1 → positive (clockwise) direction of the indexes,
ei jk =
M

.A


⎪ i, j, k ∈ {123, 231, 312}


⎪ −1 → negative (counterclockwise) direction of the indexes,
m



d

i, j, k ∈ {132, 321, 213}


uu
e
X Th

er
tin

This definition is summarized in graphic form in Figure 3.2.


on

.O
C

Figure 3.2: Definition of the permutation operator, ei jk .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
116 C HAPTER 3. C OMPATIBILITY E QUATIONS

Finally, another possible expression of the compatibility conditions is

εi j,kl + εkl,i j − εik, jl − ε jl,ik = 0 i, j, k, l ∈ {1, 2, 3} . (3.19)

Remark 3.6. Since the compatibility equations (3.16) only involve


the second spatial derivatives of the components of the strain ten-
sor ε (x,t), every strain tensor that is linear (first-order polynomial)
with respect to the spatial variables will be compatible and, there-
fore, integrable. As a particular case, every uniform strain tensor

rs
ε (t) is integrable.

ee
s gin
t d le En

r
3.4 Integration of the Infinitesimal Strain Field

ba
ge ro or
eS m
ci
3.4.1 Preliminary Equations
f

ra
C d P cs
b
Consider the rotation tensor Ω (x,t) for the infinitesimal strain case (see Chap-

a
i
an an n

ter 2 , Section 2.11.6),


y ha


⎪ 1
le
⎨ Ω = (u ⊗ ∇ − ∇ ⊗ u) ,
liv or ec

2   (3.20)
1 ∂ ui ∂ u j
M

.A


⎩ Ωi j = − i, j ∈ {1, 2, 3} .
2 ∂ x j ∂ xi
m

d
uu

and the infinitesimal rotation vector θ (x,t), associated with said rotation tensor,
e
X Th

defined as3
er
tin

⎡ ⎤ ⎡ ⎤ ⎡ ⎤
θ1 −Ω23 −Ωyz
on

.O

1 1 not ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
θ = rot u = ∇ × u ≡ ⎣ θ2 ⎦ = ⎣ −Ω31 ⎦ = ⎣ −Ωzx ⎦ . (3.21)
C

2 2
θ3 −Ω12 −Ωxy
©

Differentiating the infinitesimal rotation tensor in (3.20) with respect to a


coordinate xk yields
   
1 ∂ ui ∂ u j ∂ Ωi j 1 ∂ ∂ ui ∂ u j
Ωi j = − =⇒ = − . (3.22)
2 ∂ x j ∂ xi ∂ xk 2 ∂ xk ∂ x j ∂ xi

⎡ ⎤
0 Ω12 −Ω31
not ⎢ ⎥
3 The tensor Ω is skew-symmetric, i.e., Ω ≡ ⎣ −Ω12 0 Ω23 ⎦.
Ω31 −Ω23 0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Integration of the Infinitesimal Strain Field 117

Adding and subtracting in (3.22) the term ∂ 2 uk /(2 ∂ xi ∂ x j ) and rearranging the
expression obtained results in
 
∂ Ωi j 1 ∂ ∂ ui ∂ u j 1 ∂ 2 uk 1 ∂ 2 uk
= − + − =
∂ xk 2 ∂ xk ∂ x j ∂ xi 2 ∂ xi ∂ x j 2 ∂ xi ∂ x j
   
∂ 1 ∂ ui ∂ uk ∂ 1 ∂ u j ∂ uk ∂ εik ∂ ε jk (3.23)
= + − + = − .
∂ x j 2 ∂ xk ∂ xi ∂ xi 2 ∂ xk ∂ x j ∂xj ∂ xi
     
εik ε jk

rs
This expression can now be used to calculate the Cartesian derivatives of the

ee
components of the infinitesimal rotation vector, θ (x,t), given in (3.21), as fol-

s gin
lows. ⎧

⎪ ∂ θ1 ∂ Ωyz ∂ εxz ∂ εxy

t d le En



⎪ =− = −

⎪ ∂x ∂x ∂y ∂z

r

ba
ge ro or
eS m
∇θ1 ∂ θ1 ∂ Ωyz ∂ εyz ∂ εyy

ci
=− = − (3.24)

⎪ f

ra
⎪ ∂y ∂y ∂y ∂z
C d P cs


b
a

⎪ ∂ θ1 ∂ Ωyz ∂ εzz ∂ εzy

i

an an n

⎩ =− = −
∂z ∂z ∂y ∂z
y ha

le

liv or ec


⎪ ∂ θ2 ∂ Ωzx ∂ εxx ∂ εxz
M


.A


⎪ =− = −

⎪ ∂x ∂x ∂z ∂x


m

∇θ2 ∂ θ2 ∂ Ωzx ∂ εxy ∂ εyz


uu

=− = − (3.25)

e

⎪ ∂y
⎪ ∂y ∂z ∂x
X Th


er


tin



⎪ ∂ θ2
⎩ ∂ Ωzx ∂ εxz ∂ εzz
=− = −
on

.O

∂z ∂z ∂z ∂x

C


©


⎪ ∂ θ3 ∂ Ωxy ∂ εxy ∂ εxx

⎪ =− = −

⎪ ∂x ∂x ∂x ∂y



∇θ3 ∂ θ3 ∂ Ωxy ∂ εyy ∂ εxy (3.26)
⎪ =− = −

⎪ ∂y ∂y ∂x ∂y



⎪ ∂ Ωxy ∂ εyz ∂ εxz
⎪ ∂ θ3

⎩ =− = −
∂z ∂z ∂x ∂y

Assume the value of the infinitesimal rotation vector θ (x,t) is known and,
through it by means of (3.21), the value of the infinitesimal rotation tensor

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
118 C HAPTER 3. C OMPATIBILITY E QUATIONS

Ω (x,t) is also known. Then, the displacement gradient tensor J (x,t) (see Chap-
ter 2, Section 2.11.6) becomes

⎪ ∂ u (x,t)

⎪ J= = ε +Ω
⎨ ∂x    
∂ ui 1 ∂ ui ∂ u j 1 ∂ ui ∂ u j
⎪ Ji j = = + + − = εi j + Ω i j

⎪ ∂xj 2 ∂ x j ∂ xi 2 ∂ x j ∂ xi
⎩       i, j ∈ {1, 2, 3} .
εi j Ωi j
(3.27)
Finally, writing in explicit form the different components in (3.27) and taking

rs
into account (3.21), the following is obtained4 .

ee
s gin
j=1 j=2 j=3
∂ ux ∂ ux ∂ ux
i=1: = εxx = εxy − θ3 = εxz + θ2

t d le En
∂x ∂y ∂z
∂ uy ∂ uy ∂ uy (3.28)

r
ba
i=2: = εxy + θ3 = εyy = εyz − θ1

ge ro or
eS m
∂x ∂y ∂z

ci
∂ uz ∂ uz f ∂ uz

ra
C d P cs
i=3: = εxz − θ2 = εyz + θ1 = εzz
b
a
∂x ∂y ∂z
i
an an n
y ha

le
3.4.2 Integration of the Strain Field
liv or ec
M

.A

Consider ε (x,t) is the infinitesimal strain field one wants to integrate. This op-
eration is performed in two steps:
m

1) Using (3.24) through (3.26), the infinitesimal rotation vector θ (x,t) is in-
uu
e

tegrated. The integration, with respect to space, of the infinitesimal rotation


X Th

er
tin

vector in (3.24) through (3.26) leads to a solution of the type


on

.O

θi = θ̃i (x, y, z,t) + ci (t) i ∈ {1, 2, 3} , (3.29)


C

where the integration constants ci (t), which, in general, may be a function


©

of time, can be determined if the value (or the evolution along time) of the
infinitesimal rotation vector at some point of the medium is known.
2) Once the infinitesimal strain tensor ε (x,t) and the infinitesimal rotation
vector θ (x,t) are known, the displacement field u (x,t) is integrated. The
system of first-order PDEs defined in (3.28) is used, resulting in

ui = ũi (x, y, z,t) + ci (t) i ∈ {1, 2, 3} . (3.30)


⎡ ⎤ ⎡ ⎤
0 Ω12 −Ω31 0 −θ3 θ2
According to (3.21), Ω ≡ ⎣ −Ω12 0 Ω23 ⎦ = ⎣ θ3 0 −θ1 ⎦.
4 not

Ω31 −Ω23 0 −θ2 θ1 0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Integration of the Infinitesimal Strain Field 119

Again, the integration constants ci (t) that appear, which, in general, will be
a function of time, are determined when the value (or the evolution along
time) of the displacements at some point of space is known.

Remark 3.7. The integration processes in steps 1) and 2) involve in-


tegrating systems of first-order PDEs. If the compatibility equations
in (3.16) are satisfied, these systems will be integrable (without lead-
ing to contradictions in their integration process) and will finally al-
low obtaining the displacement field.

rs
ee
s gin
Remark 3.8. The presence of the integration constants in (3.29) and

t d le En
(3.30) shows that an integrable strain tensor, ε (x,t), determines the

r
not
motion of each instant of time except for a rotation c (t) = θ̂θ (t) and

ba
ge ro or
eS m
 not
a translation c (t) = û (t).

ci
 f

ra
θ (x,t) = θ̃θ (x,t) + θ̂θ (t)
C d P cs
ε (x,t)
b
a
u (x,t) = ũ (x,t) + û (t)
i
an an n
y ha

From these uniform rotation θ̂θ (t) and translation û (t) the displace-
le
liv or ec

ment field
M

u∗ (x,t) = Ω̂ u∗ ⊗ ∇ = Ω̂
.A

Ω (t) x + û (t) =⇒ Ω
m

can be defined, which corresponds to a rigid body motion5 . Indeed,


uu
e

the strain associated with this displacement is null,


X Th

er
tin

1 1 
ε ∗ (x,t) = ∇s u∗ = (u∗ ⊗ ∇ + ∇ ⊗ u∗ ) = Ω̂ Ω + 
ΩT = 0 ,
Ω̂
2 2
on

.O

Ω
−Ω̂
C

as corresponds to the concept of rigid body (without deformation).


©

Consequently, it is concluded that every compatible strain field de-


termines the displacements of the continuous medium except for a
rigid body motion, which must be determined by means of the ap-
propriate boundary conditions.

5 Ω (t) (antisymmetric) is defined based on the rotation vector


The rigid body rotation tensor Ω̂
⎡ ⎤ ⎡ ⎤
0 Ω̂12 −Ω̂31 0 −θ̂3 θ2
Ω ≡ ⎣ −Ω̂12 0 Ω̂23 ⎦ = ⎣ θ3 0 −θ1 ⎦ .
not
θ̂θ (t) as Ω̂
Ω̂31 −Ω̂23 0 −θ2 θ1 0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
120 C HAPTER 3. C OMPATIBILITY E QUATIONS

Example 3.1 – A certain motion is defined by the infinitesimal strain tensor


⎡ ⎤
y 3 2
⎢ 8x − x z
2 2 ⎥
not ⎢ y ⎥
ε (x,t) ≡ ⎢ −
⎢ 2 x 0 ⎥.

⎣3 ⎦
2
x z 0 x 3
2
Obtain the corresponding displacement vector u (x,t) and the infinitesimal
rotation tensor Ω (x,t) taking into account that u (x,t)|x=[0,0,0]T ≡ [3t, 0, 0]T
not

rs
and Ω (x,t)|x=[0,0,0]T = 0.

ee
s gin
Solution

t d le En
Infinitesimal rotation vector

r
Posing the systems of equations defined in (3.24) through (3.26) results in

ba
ge ro or
eS m
ci
∂ θ1 ∂ θ1 ∂ θ1
f

ra
=0 ; =0 ; =0 ⇒ θ1 = C1 (t) ,
C d P cs
∂x ∂y ∂z
b
a
i
∂ θ2 ∂ θ2 ∂ θ2
an an n

3 3
= −3xz ; =0 ; = − x2 ⇒ θ2 = − x2 z +C2 (t) ,
y ha

∂x ∂y ∂z 2 2
le
∂ θ3 ∂ θ3 3 ∂ θ3
liv or ec

3
=0 ; = ; =0 ⇒ θ3 = y +C3 (t) .
∂x ∂y ∂z
M

2 2
.A

The integration constants Ci (t) are determined by imposing that


m

Ω (x,t)|x=(0,0,0)T = 0 (and, therefore, the infinitesimal rotation vector


uu
e

θ (x,t)|x=(0,0,0)T = 0), that is,


X Th

er
tin

 T
3 3
on

.O

not
C1 (t) = C2 (t) = C3 (t) = 0 =⇒ θ (x) ≡ 0 , − x2 z , y
2 2
C

and the infinitesimal rotation tensor is


⎡ ⎤
3 3 2
⎡ ⎤ ⎢ 0 − y − x z⎥
0 −θ3 θ2 ⎢ 2 2 ⎥
⎢ ⎥
not ⎢ ⎥ ⎢ ⎥
Ω (x) ≡ ⎣ θ3 0 −θ1 ⎦ = ⎢ 3 y 0 0 ⎥ .
⎢ 2 ⎥
−θ2 θ1 0 ⎢ ⎥
⎣3 ⎦
x2 z 0 0
2

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Compatibility Equations and Integration of the Strain Rate Field 121

Displacement vector
Posing, and integrating, the systems of equations in (3.28) produces

∂ u1 ∂ u1 ∂ u1
= 8x ; = −2y ; =0 ⇒ u1 = 4x2 − y2 +C1 (t) ,
∂x ∂y ∂z
∂ u2 ∂ u2 ∂ u2
=y ; =x ; =0 ⇒ u2 = xy +C2 (t) ,
∂x ∂y ∂z
∂ u3 ∂ u3 ∂ u3
= 3x2 z ; =0 ; = x3 ⇒ u3 = x3 z +C3 (t) .

rs
∂x ∂y ∂z

ee
and imposing that u (x,t)|x=(0,0,0)T ≡ [3t, 0, 0]T yields
not

s gin
not  T

t d le En
C1 (t) = 3t ; C2 (t) = C3 (t) = 0 =⇒ u (x) ≡ 4x2 − y2 + 3t , xy , x3 z .

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
3.5 Compatibility Equations and Integration of the Strain
i
an an n

Rate Field
y ha

le
Given the definitions of the infinitesimal strain tensor ε , the infinitesimal rotation
liv or ec

tensor Ω and the infinitesimal rotation vector θ , there exists a clear correspon-
M

.A

dence between these magnitudes and a) the strain rate tensor d, b) the rotation
rate (or spin) tensor w and c) the spin vector ω given in Chapter 2. These corre-
m

spondences can be established in the following manner:


uu
e
X Th

er
tin

u v
on

.O

ε (u) d (v)
   
C

1 ∂ ui ∂ u j 1 ∂ vi ∂ v j
©

εi j = + di j = +
2 ∂ x j ∂ xi 2 ∂ x j ∂ xi
⇐⇒ (3.31)
   
1 ∂ ui ∂ u j 1 ∂ vi ∂ v j
Ωi j = − wi j = −
2 ∂ x j ∂ xi 2 ∂ x j ∂ xi
1 1
θ = ∇×u ω = ∇×v
2 2

Then, it is obvious that the concept of compatibility of a strain field ε in-


troduced in Section 3.1 can be extended, by virtue of the correspondence with
(3.31), to the compatibility of a strain rate field d (x,t).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
122 C HAPTER 3. C OMPATIBILITY E QUATIONS

To integrate this field, the same procedure as that seen in Section 3.4.2 can be
used, replacing ε by d, u by v, Ω by w and θ by ω . Certainly, this integration
can only be performed if the compatibility equations in (3.16) are satisfied for
the components of d (x,t).

Remark 3.9. The resulting compatibility equations and the integra-


tion process of the strain rate vector d (x,t) are not, in this case,
restricted to the infinitesimal strain case.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 123

P ROBLEMS

Problem 3.1 – Determine the spatial description of the velocity field that cor-
responds to the strain rate tensor
⎡ ⎤
tetx 0 0
not ⎢ ⎥

rs
d (x,t) ≡ ⎢ ⎣ 0 0 te y +1⎥ .

ee
0 te + 1
y 0

s gin
For x = 0, ω 0 ≡ [t − 1, 0, 0]T and v0 ≡ [t, 0, t]T f or ∀t is satisfied.
not not

t d le En

r
ba
ge ro or
eS m
ci
f

ra
Solution
C d P cs
b
a
The problem is solved by integrating the corresponding differential equations,
i
an an n

taking into account the existent parallelism between the variables:


y ha

⎫ ⎧
le
liv or ec

u⎪⎬ ⎪
⎨v
M

.A

ε ⇐⇒ d

⎭ ⎪

θ ω
m

d
uu
e

Angular velocity of the rotation vector


X Th

er
tin

∂ ω1 ∂ ω1 ∂ ω1
on

=0 ; = tey ; =0 ⇒ ω1 = C1 (t) + tey ,


.O

∂x ∂y ∂z
C

∂ ω2 ∂ ω2 ∂ ω2
©

=0 ; =0 ; =0 ⇒ ω2 = C2 (t) ,
∂x ∂y ∂z
∂ ω3 ∂ ω3 ∂ ω3
=0 ; =0 ; =0 ⇒ ω3 = C3 (t) .
∂x ∂y ∂z
The boundary conditions are imposed for x = 0,
⎡ ⎤ ⎡ ⎤ ⎧
t −1 t +C1 ⎨ C1 = −1

not ⎢ ⎥ ⎢ ⎥
ω 0 ≡ ⎣ 0 ⎦ = ⎣ C2 ⎦ =⇒ C2 = 0 ,


0 C3 C3 = 0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
124 C HAPTER 3. C OMPATIBILITY E QUATIONS

and the final result is ⎡ ⎤


tey − 1
not ⎢ ⎥
ω (x,t) ≡ ⎣ 0 ⎦.
0
Velocity vector

∂ v1 ∂ v1 ∂ v1 
= tetx ; =0 ; =0 ⇒ v1 = C1 (t) + etx ,
∂x ∂y ∂z

rs
∂ v2 ∂ v2 ∂ v2 
=0 ; =0 ; =2 ⇒ v2 = C2 (t) + 2z ,

ee
∂x ∂y ∂z

s gin
∂ v3 ∂ v3 ∂ v3 
=0 ; = 2tey ; =0 ⇒ v3 = C3 (t) + 2tey .
∂x ∂y ∂z

t d le En
The boundary conditions are imposed for x = 0,

r
ba
ge ro or
⎡ ⎤ ⎡  ⎤ ⎧ 

eS m
ci
t 1 +C1 ⎨ C1 = t − 1

f

ra
not ⎢ ⎥ ⎢  ⎥ 
C d P cs
v0 ≡ ⎣ 0 ⎦ = ⎣ C2 ⎦ =⇒ C2 = 0 ,
b
a

⎩ 
i

2t +C3 C3 = −t
an an n

t
y ha

le
and the spatial description of the velocity field is
liv or ec

⎡ ⎤
M

.A

etx + t − 1
not ⎢ ⎥
m

v (x) ≡ ⎣ 2z ⎦ .
d
uu
e

2tey − t
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 125

E XERCISES

3.1 – Deduce the displacement field that corresponds to the infinitesimal strain
tensor ⎡ ⎤
0 tety 0
not ⎢ ty ⎥
ε (x,t) ≡ ⎢ ⎣ te 0 0 ⎥.

0 0 tetz

rs
ee
At point (1, 1, 1) , u ≡ [2et , et , et ]T and θ ≡ [0, 0, −tet ]T is verified.
not not

s gin
3.2 – Determine the spatial description of the velocity field that corresponds to

t d le En
the strain rate tensor
⎡ ⎤

r
ba
tetz

ge ro or
0 0

eS m
not ⎢ ⎥

ci
d (x,t) ≡ ⎢ ty
f ⎥.

ra
⎣ 0 te 0 ⎦
C d P cs
b
a
tetz 0 0
i
an an n
y ha

The following is known:


le
liv or ec


for z = 0 : vx = vz = 0 , ∀t , x, y
M

.A

for y = 1 : vy = 0 , ∀t , x, z
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
CH.4. STRESS
Multimedia Course on Continuum Mechanics
Overview
 Forces Acting on a Continuum Body Lecture 1
 Cauchy’s Postulates Lecture 2
 Stress Tensor Lecture 3 Lecture 4
 Stress Tensor Components
 Scientific Notation
Lecture 5
 Engineering Notation
 Sign Criterion
 Properties of the Cauchy Stress Tensor
 Cauchy’s Equation of Motion Lecture 6
 Principal Stresses and Principal Stress Directions
 Mean Stress and Mean Pressure
Lecture 7
 Spherical and Deviatoric Parts of a Stress Tensor
 Stress Invariants

2
Overview (cont’d)
 Stress Tensor in Different Coordinate Systems
 Cylindrical Coordinate System Lecture 8
 Spherical Coordinate System
 Mohr’s Circle Lecture 9
 Mohr’s Circle for a 3D State of Stress
Lecture 10
 Determination of the Mohr’s Circle
 Mohr’s Circle for a 2D State of Stress
 2D State of Stress Lecture 11
 Stresses in Oblique Plane
 Direct Problem
Lecture 12
 Inverse Problem
 Mohr´s Circle for a 2D State of Stress Lecture 13

3
Overview (cont’d)
 Mohr’s Circle a 2D State of Stress (cont’d)
 Construction of Mohr’s Circle
 Mohr´s Circle Properties Lecture 14
 The Pole or the Origin of Planes Lecture 15
 Sign Convention in Soil Mechanics Lecture 16
 Particular Cases of Mohr’s Circle Lecture 17

4
4.1. Forces on a Continuum Body
Ch.4. Stress

5
Forces Acting on a Continuum Body
Forces acting on a continuum body:
 Body forces.
 Act on the elements of volume or mass inside the body.
 “Action-at-a-distance” force.
 E.g.: gravity, electrostatic forces, magnetic forces
fV = ∫ ρ b ( x, t ) dV body force per unit
V
mass
 Surface forces. (specific body forces)

 Contact forces acting on the body at its boundary surface.


 E.g.: contact forces between bodies, applied point or distributed
loads on the surface of a body
f S = ∫ t ( x, t ) dS surface force
∂V (traction vector)
per unit surface

6
4.2. Cauchy’s Postulates
Ch.4. Stress

7
Cauchy’s Postulates
1. Cauchy’s 1st postulate. REMARK
The traction vector t remains unchanged The traction vector (generalized to
for all surfaces passing through the point P internal points) is not influenced by
and having the same normal vector n at P . the curvature of the internal surfaces.

t = t ( P, n )

2. Cauchy’s fundamental lemma


(Cauchy reciprocal theorem)
The traction vectors acting at point P
on opposite sides of the same surface
are equal in magnitude and opposite
in direction.
t ( P, n ) =
−t ( P, −n ) REMARK
Cauchy’s fundamental lemma is equivalent to
Newton's 3rd law (action and reaction).

8
4.3. Stress Tensor
Ch.4. Stress

9
Stress Tensor
 The areas of the faces of the tetrahedron
are:
S1 = n1S
n ≡ {n 1 , n 2 , n 3 }
T
S 2 = n2 S with
S3 = n3 S

 The “mean” stress vectors acting on these faces are


t* = t (x*S , n), − t (1)* = t (x*S , −eˆ1 ), − t ( 2)* = t (x*S , −eˆ 2 ), − t ( ) = t (x*S3 , −eˆ 3 )
3*

 *
1 2

x S= ∈ S ∈ S → mean value theorem


*
i
S i i 1, 2,3 ; x

 The surface normal vectors of the planes perpendicular to the axes are
n1 = −eˆ1 ; n2 = −eˆ 2 ; n3 = −eˆ 3 REMARK
 Following Cauchy’s fundamental lemma: The asterisk indicates an
t ( x, −eˆ i ) =
not
−t ( x, eˆ i ) =
− t (i ) ( x ) i ∈ {1, 2,3} mean value over the area.

10
Mean Value Theorem
 Let f : [ a, b ] → R be a continuous function on the closed interval
[a, b] , and differentiable on the open interval ( a, b ) , where a < b .
Then, there exists some x* in ( a, b ) such that:

f ( x* )
1
f ( x ) dΩ
Ω Ω∫
=

 I.e.: f : [ a, b ] → R gets its


“mean value” f ( x* ) at the interior
of [ a, b ]

11
Stress Tensor
 From equilibrium of forces, i.e. Newton’s 2nd law of motion:
R=∑ fi =∑ mi ai ∫ ρ b dV + ∫ t dS =∫ a ρ ∫ ρ a dV
dV =
i i V ∂V V dm V
resultant
body forces
(1) ( 2) ( 3)

V
ρ b dV + ∫ t
S
dS + ∫ − t
S1
dS + ∫ − t
S2
dS + ∫ dS =∫ ρ a dV
− t
S3 V
resultant
 Considering the mean value theorem, surface forces

( ρ b)* V + t*S − t (1)*S1 − t ( 2)*S2 − t (3)*S3 =


( ρ a)* V

1
 Si ni S i ∈ {1, 2,3} and V = Sh ,
Introducing=
3
1 1
( ρ b)* h S + t*S − t (1)*n1S − t ( 2)*n2S − t (3)*n3S = ( ρ a)* hS
3 3

12
Stress Tensor
 If the tetrahedron shrinks to point O,

x*Si → xO (
lim t (i )* x*Si , eˆ i  =
h →0   )
t (i ) ( O,eˆ i ) i ∈ {1, 2,3}

x*S → xO lim t* ( x*S , n )  =


t ( O, n )
h →0

1  1 
lim= ( ρ b )*
h =
 h →0 
lim ( ρ a )*
h  0

h →0 3
 3 

 The limit of the expression for the equilibrium of forces becomes,

= t (1) = t ( 2) = t ( 3)
1 (1)* ( 2 )* ( 3)* 1
( ρ b) h + t − t n1 − t n2 − t n3 =( ρ a)* h
* *
t ( O, n ) − t (i ) ni =
0
3 3
= t ( O, n )

13
Stress Tensor
 Considering the traction vector’s Cartesian components :
=t (i ) ( P ) t =
(i )
ˆ σ ij eˆ j
j ( P) e j t ( P=
, n ) t (i ) ni ⇒
 i, j ∈ {1, 2,3}
σ ij ( P ) = t j ( P )
(i )
t j ( P=
,n) t=
(i )
j ni niσ ij
 
 σ ij
Cauchy’s Stress Tensor t ( P, n )= n ⋅ σ ( P)

σ σ ij eˆ i ⊗ eˆ j
=
P

 In the matrix form:


t1(1) t1( 2) t1( 3)
t j n=
= i σ ij σ Tji ni

 j ∈ {1, 2,3}

[ t ] = [ σ ] [n ]
T

t (1) t ( 2) t ( 3)
14
Stress Tensor
REMARK 1
The expression t ( P, n )= n ⋅ σ ( P ) is consistent with Cauchy’s postulates:
t ( P, n )= n ⋅ σ
t ( P, n ) =
−t ( P, −n )
t ( P, −n ) =−n ⋅ σ

REMARK 2
The Cauchy stress tensor is constructed from the traction vectors on three
coordinate planes passing through point P.

σ 11 σ 12 σ 13 
σ ≡ σ 21 σ 22 σ 23 
σ 31 σ 32 σ 33 

Yet, this tensor contains information on the traction vectors acting on any plane
(identified by its normal n) which passes through point P.

15
4.4.Stress Tensor Components
Ch.4. Stress

16
Scientific Notation
 Cauchy’s stress tensor in scientific notation

σ 11 σ 12 σ 13 
σ ≡ σ 21 σ 22 σ 23 
σ 31 σ 32 σ 33 

 Each component σ ij is characterized by its sub-indices:


 Index i designates the coordinate plane on which the component acts.
 Index j identifies the coordinate direction in which the component acts.

17
Engineering Notation
 Cauchy’s stress tensor in engineering notation

σ x τ xy τ xz 
 
σ ≡ τ yx σ y τ yz 
τ zx τ zy σ z 
 

 Where:
 σ a is the normal stress acting on plane a.
 τ ab is the tangential (shear) stress acting on the plane perpendicular to
the a-axis in the direction of the b-axis.

18
Tension and compression
 The stress vector acting on point P of an
arbitrary plane may be resolved into:
 a vector normal to the plane (σ n = σ n)
 an in-plane (shear) component which acts on the plane.
(τ n ; τ n = τ )
 The sense of σ n with respect to n defines the normal stress character:
σ= σ n ⋅n >0 tensile stress (tension)
<0 compressive stress (compression)
 The sign criterion for the stress components is:
positive (+) tensile stress
σ ij or σ a negative (−) compressive stress
positive (+) positive direction of the b-axis
τ ab negative (−) negative direction of the b-axis
19
4.5.Properties of the Cauchy Stress
Tensor
Ch.4. Stress

20
Cauchy’s Equation of Motion
 Consider an arbitrary material volume,
 Cauchy’s equation of motion is:
∇ ⋅ σ + ρ= b ρ a ∀x ∈ V

 ∂σ ij
 ∂x + ρ b=
j ρ a j j ∈ {1, 2,3}
 i
 In engineering notation: b ( x, t ) x ∈ V
∂σ x ∂τ yx ∂τ zx t* ( x, t ) x ∈ ∂V
+ + + ρ bx =ρ ax
∂x ∂y ∂z
∂τ xy ∂σ y ∂τ zy REMARK
+ + + ρ by =ρ ay
∂x ∂y ∂z Cauchy’s equation of motion is derived
∂τ xz ∂τ yz ∂σ z from the principle of balance of linear
+ + + ρ bz =ρ az momentum.
∂x ∂y ∂z

21
Equilibrium Equations
 For a body in equilibrium a = 0 ,
Cauchy’s equation of motion becomes
∇ ⋅ σ + ρ b = 0 ∀x ∈ V
 ∂σ internal equilibrium
 ij + ρ b = 0 j ∈ {1, 2,3}
 ∂xi j equation

 The traction vector is now known at


the boundary
n ( x, t ) ⋅ σ=
( x, t ) t* ( x, t ) ∀x ∈ ∂V equilibrium equation

=
ni σij t *j j ∈ {1, 2,3} at the boundary

 The stress tensor symmetry is derived from the principle of balance of


angular momentum:
σ = σ T
σ i, j ∈ {1, 2,3}
=  ij σ ji

22
Cauchy’s Equation of Motion
 Taking into account the symmetry of the
Cauchy Stress Tensor,
 Cauchy’s equation of motion
∇ ⋅ σ + ρ b = σ ⋅ ∇ + ρ b = ρ a ∀x ∈ V

 ∂σ ij ∂σ ji
 ∂x + ρ b = + ρbj = ρ a j j ∈ {1, 2,3}

j
 i xi

b ( x, t ) x ∈ V
 Boundary conditions
t* ( x, t ) x ∈ ∂V
n ⋅ σ= σ ⋅ n= t (x, t ) ∀x ∈ ∂V
*


ni σij = σ ji ni = t j ( x, t ) ∀x ∈ ∂V i, j ∈ {1, 2,3}
*

23
Principal Stresses and Principal
Stress Directions
 Regardless of the state of stress, it is always possible to choose a
special set of axes (principal axes of stress or principal stress
directions) so that the shear stress components vanish when the
stress components are referred to this system.
 The three planes perpendicular to the principal axes are the
principal planes.
 The normal stress components in the principal planes are the
principal stresses. σ 33
x3 σ 31 σ 32 x3
σ 13 x3′
σ 1 0 0 
σ 23
σ 11 σ 12 σ 21 σ x1′ σ3
[σ ] =  0 σ 2 0  22
σ1
 0 0 σ 3 
σ2
x1 x1
x2 x2
x2′
24
Principal Stresses and Principal
Stress Directions
 The Cauchy stress tensor is a symmetric 2nd order tensor so it will
diagonalize in an orthonormal basis and its eigenvalues are
real numbers.
 For the eigenvalue λ and its corresponding eigenvector v :
σ⋅v =λv [ σ − λ 1] ⋅ v =0 INVARIANTS
not characteristic
det [ σ − λ 1] = σ − λ 1 =0 λ 3 − I 1 ( σ )λ 2 − I 2 ( σ )λ − I 3 ( σ ) =
0 equation
σ 33
λ1 ≡ σ 1 x3 σ 31 σ 32 x3
x3′
λ2 ≡ σ 2 σ 13 σ 23
λ3 ≡ σ 3 σ 11 σ 12 σ 21 σ x1′ σ3
σ1
22

REMARK σ2
x1 x1
The invariants associated with a x2 x2
tensor are values which do not change x2′

25
with the coordinate system being used.
Mean Stress and Mean Pressure
 Given the Cauchy stress tensor σ and its principal stresses, the
following is defined:
σ 11 σ 12 σ 13 
 Mean stress
σ ≡ σ 21 σ 22 σ 23 
1 1 1
σ m= Tr (σ )= σ ii = (σ 1 + σ 2 + σ 3 ) σ 31 σ 32 σ 33 
3 3 3
 Mean pressure
REMARK
1 In a hydrostatic state of stress, the
− (σ 1 + σ 2 + σ 3 )
−σ m =
p=
3 stress tensor is isotropic and, thus,
its components are the same in
 A spherical or hydrostatic any Cartesian coordinate system.
state of stress: σ 0 0 As a consequence, any direction
is a principal direction and the
σ= σ= σ3 σ ≡  0 σ 0  =
σ1
1 2
stress state (traction vector) is the
 0 0 σ  same in any plane.

26
Spherical and Deviatoric Parts of a
Stress Tensor
 The Cauchy stress tensor σ can be split into:
σ = σ sph + σ ′
 The spherical stress tensor:
 Also named mean hydrostatic stress tensor or volumetric stress tensor or
mean normal stress tensor.
 Is an isotropic tensor and defines a hydrostatic state of stress.
 Tends to change the volume of the stressed body
1 1
σ=
sph : σ =
m 1 Tr ( σ=) 1 σ ii 1 REMARK
3 3 The principal directions of a stress tensor
 The stress deviator tensor: and its deviator stress component coincide.
 Is an indicator of how far from a hydrostatic state of stress the state is.
 Tends to distort the volume of the stressed body
σ ′ dev
= = σ σ − σ m1

27
Stress Invariants
 Principal stresses are invariants of the stress state:
 invariant w.r.t. rotation of the coordinate axes to which the stresses are
referred.
 The principal stresses are combined to form the stress invariants I :
I1 = Tr ( σ ) = σ ii = σ 1 + σ 2 + σ 3 REMARK
I 2 =( σ : σ − I12 ) =
1 The I invariants are obtained
− (σ 1σ 2 + σ 1σ 3 + σ 2σ 3 )
2 from the characteristic equation
I 3 = det ( σ ) of the eigenvalue problem.
 These invariants are combined, in turn, to obtain the invariants J :
J=
1 I=
1 σ ii REMARK
The J invariants can be
J2 =
2
(
1 2
I1 + 2 I 2 ) = σ ijσ ji = ( σ : σ )
1
2
1
2 expressed in the unified form:
J=
3
3
(
1 3
I1 + 3 I I
1 2 + 3 I =
3 ) 1
3
Tr ( σ ⋅ σ ⋅ σ
= )
1=
3
σ σ σ
ij jk ki
J i
1
i
Tr( )
σ i
i ∈ {1, 2,3}

28
Stress Invariants of the Stress
Deviator Tensor
 The stress invariants of the stress deviator tensor:
= I1′ Tr
= ( σ′ ) 0
=I 2′
1
2
( )
σ′ : σ′ − I12 = σ 12′ σ 12′ + σ 13′ σ 13′ + σ 23
′ σ 23′

′ = (σ ij′σ ′jkσ ki′ )


1
I 3′ = det ( σ′ ) = σ 11′ σ 22
′ σ 33
′ + 2σ 12′ σ 23
′ σ 13′ − σ 12′2σ 33
′ − σ 23
′2σ 11′ − σ 13′2σ 22
3
 These correspond exactly with the invariants J of the same stress
deviator tensor:
′ I=
J=
1
′ 0
1

J 2′ =
1 2
2
( )
I1′ + 2 I 2′ =
1
I 2′ = ( σ′ : σ′ )
2

J 3′ =
3
( )
I1′ + 3I1′I 2′ + 3I 3′ = I 3′ = Tr ( σ′ ⋅ σ′ ⋅ σ′ ) = (σ ij′σ ′jkσ ki′ )
1 3 1
3
1
3

29
4.6. Stress Tensor in Different
Coordinate Systems
Ch.4. Stress

30
Stress Tensor in a Cylindrical
Coordinate System
 The cylindrical coordinate system is defined by:
dV = r dθ dr dz

 x = r cos θ

x( r , θ , z ) ≡  y = r sin θ
z = z

 The components of the stress tensor are then:
 σ x´ τ x´ y ´ τ x´ z ´  σ r τ rθ τ rz 
  τ
=σ =
τ x´ y ′ σ y ´ τ y ´ z ´   rθ σ θ τ θ z 
τ x´ z´ τ y´ z´ σ z´  τ rz τ θ z σ z 
 

31
Stress Tensor in a Spherical
Coordinate System
dV = r 2 sen θ dr dθ dϕ
 The cylindrical coordinate system is defined by:

 x = r sen θ cos φ

x ( r ,θ , ϕ ) ≡  y = r sen θ sen φ
 The components of the stress tensor are then:  z = r cos θ

 σ x´ τ x´ y´ τ x´ z´  σ r τ rθ τ rφ 
   
σ ≡ τ x´ y′ σ y´ τ y´ z´  = τ
 rθ σ θ τ θφ 
τ x´ z´ τ y´ z´ σ z´  τ rφ τ φθ σ φ 
   

32
4.7. Mohr´s Circle
Ch.4. Stress

33
Mohr’s Circle
 Introduced by Otto Mohr in 1882.
 Mohr´s Circle is a two-dimensional graphical representation of
the state of stress at a point that:
 will differ in form for a state of stress in 2D or 3D.
 illustrates principal stresses and maximum shear stresses as well as stress
transformations.
 is a useful tool to rapidly grasp
the relation between stresses for a
given state of stress.

34
4.8. Mohr´s Circle for a 3D State of
Stress
Ch.4. Stress

35
Determination of Mohr’s Circle
 Consider the system of Cartesian axes linked to the principal
directions of the stress tensor at an arbitrary point P of a
continuous medium: x3
ê3
 The components of the stress tensor are ê1 σ3
σ 1 0 0  x1
σ ≡  0 σ 2 0  with σ2
 0 0 σ 3  ê2
x2
 The components of the traction vector are
σ 1 0 0   n1   σ 1 n1 
t = σ ⋅ n =  0 σ 2 0   n2  = σ n 
 2 2
 0 0 σ 3   n3  σ 3 n3 

where n is the unit normal to the base associated to the principal directions

36
Determination of Mohr’s Circle
 The normal component of stress σ is
 n1 
σ = t ⋅ n = [σ 1n1 , σ 2 n2 , σ 3n3 ]  n2  = σ 1n12 + σ 2 n22 + σ 3n32

T  n3 
t  
 
n  σ n= σ ⋅n
 
 The squared modulus of the traction vector is

t = t ⋅ t = σ 12n12 + σ 22n22 + σ 32n32 


2

 σ 12n12 + σ 22n22 + σ 32n32 =σ 2 +τ 2


t = σ + τ : τ := τn
2

2 2

Mohr's 3D problem half - space


 The unit vector n must satisfy
n =1 n12 + n22 + n32 =
1

 Locus of all possible (σ ,τ ) points?

37
Determination of Mohr’s Circle
 The previous system of equations can be written as a matrix
equation which can be solved for any couple
σ 12 σ 22 σ 32   n12  σ 2 + τ 2 
   2  
σ
 1 σ 2 σ 3 n
 2  = σ 
1 1 1   n3   1 
2
      
A x b
0 ≤ n12 ≤ 1
A feasible solution for x ≡  n12 , n22 , n32  requires that  for the
T

0 ≤ n2 ≤ 1
2

expression n12 + n22 + n32 =


1 to hold true. 0 ≤ n 2 ≤ 1
 3

 Every couple of numbers (σ ,τ ) which leads to a solution x , will be


considered a feasible point of the half-space.
 The feasible point is representative of the traction vector (σ ,τ ) on a
T
plane of normal n ≡  n1 , n2 , n3  which passes through point P.
 The locus of all feasible points is called the feasible region.

38
Determination of Mohr’s Circle
 The system σ 12 σ 22 σ 32   n12  σ 2 + τ 2 
   2  
σ
 1 σ 2 σ 3 n
 2  = σ 
1 1 1   n3   1 
2
      
A x b

can be re-written as
( I ) → σ 2 + τ 2 − (σ 1 + σ 3 ) σ + σ 1σ 3 −
A
n12 =
0
(σ 1 − σ 3 )
( II ) → σ 2 + τ 2 − (σ 2 + σ 3 ) σ + σ 2σ 3 −
A
n22 =
0
(σ 2 − σ 3 )
( III ) → σ 2 + τ 2 − (σ 1 + σ 2 ) σ + σ 1σ 2 −
A
n32 =
0
(σ 1 − σ 2 )
with (σ 1 − σ 2 ) (σ 2 − σ 3 ) (σ 1 − σ 3 )
A=

39
Determination of Mohr’s Circle
 Consider now equation ( III ) :
σ 2 + τ 2 − (σ 1 + σ 2 ) σ + σ 1σ 2 −
A
0 with
n32 = (σ 1 − σ 2 ) (σ 2 − σ 3 ) (σ 1 − σ 3 )
A=
(σ 1 − σ 2 )
 It can be written as: = a
1
(σ 1 + σ 2 )
2
(σ − a ) + τ 2 =
2
R 2 with 1
(σ 1 − σ 2 ) + (σ 2 − σ 3 ) (σ 1 − σ 3 ) n32
2
=
R
4
which is the equation of a semicircle of center C3 and radius R3 :
1 
=C3  (σ 1 + σ 2 ) , 0 
2  REMARK
A set of concentric semi-circles is
1
( σ 1 − σ 2 ) + (σ 2 − σ 3 ) (σ 1 − σ 3 ) n32
2
R=
3
4 obtained with the different values of
n3 with center C3 and radius R3 ( n3 ) :
n32 = 0 =
R3min
1
2
(σ 1 − σ 2 )
n32 = 1 R3max =
1
2
(σ 1 + σ 2 ) − σ 3

40
Determination of Mohr’s Circle
 Following a similar procedure with ( I ) and ( II ) , a total of three
semi-annuli with the following centers and radii are obtained:
1
1
C1 [ (σ 2 + σ 3 ) , 0]
=
=
R1min (σ 2 − σ 3 )
2
2

a1 = σ 1 − a1
R1max

1
1
C2 [ (σ 1 + σ 3 ) , 0]
= =
R2
max
(σ 1 − σ 3 )
2 2

a2 = σ 2 − a2
R2min

1 1
=
C3 [ (σ 1 + σ 2 ) , 0] =
R3min (σ 1 − σ 2 )
2
 2
a3 = σ 3 − a3
R3max

41
Determination of Mohr’s Circle
 Superposing the three annuli,

 The final feasible region must be the intersection of these semi-annuli


 Every point of the feasible region in the Mohr’s space, corresponds to
the stress (traction vector) state on a certain plane at the considered
point
44
4.9. Mohr´s Circle for a 2D State of
Stress
Ch.4. Stress

45
2D State of Stress
3D general state of stress 2D state of stress
σ x τ xy τ xz  σ x τ xy 0 
  σ ≡ τ yx σ y 0 
σ ≡ τ yx σ y τ yz 
τ zx τ zy σ z   0 0 σ z 
 

3D problem

REMARK σ x τ xy 
σ≡ 
In 2D state of stress problems, the τ yx σ y 
principal stress in the disregarded
direction is known (or assumed) a priori. 2D (plane)
problem

46
Stresses in a oblique plane
 Given a plane whose unit normal n forms an angle θ with the
x axis,
 Traction vector
σ x τ xy  cos θ  σ x cos θ + τ xy sin θ 
t = σ ⋅n =    = τ cos θ + σ sin θ 
τ
  σ θ
  xy 
sin
xy y
 y

σ  n 
   

 Normal stress
σ x +σ y σ x −σ y
σ θ =t ⋅ n = + cos ( 2θ ) + τ xy sin ( 2θ ) cos θ 
=
 sin θ 
2 2 n=   m  − cos θ 
sin θ   
 Shear stress
σ x −σ y
τ θ =t ⋅ m = sin ( 2θ ) − τ xy cos ( 2θ )
2

 Tangential stress τ θ is now endowed with sign (τ θ ≥ 0 or τ θ < 0)


 Pay attention to the “positive” senses given in the figure
47
Direct and Inverse Problems
 Direct Problem: Find the principal
stresses and principal stress directions
given σ in a certain set of axes.
equivalent stresses
 Inverse Problem: Find the stress state
on any plane, given the principal σ
stresses and principal stress directions.

48
Direct Problem
 In the x´and y´axes, τ α = 0 then,
σ x −σ y
τα
= sin ( 2α ) − τ xy cos ( 2=
α) 0
2
τ xy
tan ( 2α ) =
σ x −σ y
2
 Using known trigonometric relations,
1 τ xy
sin ( 2α ) =
± =
±
This equation has two solutions:
1  σ x −σ y 
2
1+ 2
tg ( 2α )  + τ xy 1. α1 ( sign "+ ")
2

 2 
π
σ x −σ y 2. α2 =
α1 + ( sign " − ")
2
1 These define the principal stress
cos ( 2α ) =
± =
± 2
1 + tg ( 2α )
2
 σ x −σ y 
2
directions.
 + τ xy
2

 2  (The third direction is perpendicular to
the plane of analysis.)
49
Direct Problem
 The angles θ = α1 and θ = α 2 are then introduced into the equation
σ x +σ y σ x −σ y
σθ = + cos ( 2θ ) + τ xy sin ( 2θ )
2 2

to obtain the principal stresses


(orthogonal to the plane of analysis):

 σx +σ y  σ x −σ y 
2
σ 1 = +   + τ xy
2

 2  2 
σα → 
 σx +σ y  σ x −σ y 
2

σ 2 = −  + τ xy
2

 2  2  θ ≡α

50
Inverse Problem
 Given the directions and principal stresses σ 1 and σ 2, to find the
stresses in a plane characterized by the angle β :
 Take the equations

 Replace , , and θ ≡ β
to obtain:
σ1 + σ 2 σ1 − σ 2
σβ
= + cos ( 2 β )
2 2
σ −σ
τ β = 1 2 sin ( 2β )
2

51
Mohr’s Circle for a 2D State of Stress

 Considering a reference system x´− y´


and characterizing the inclination of a
plane by ,
 From the inverse problem equations:
σ1 + σ 2 σ1 − σ 2
σ− = cos ( 2 β )
2 2
σ −σ
τ = 1 2 sin ( 2β )
2 σ1 − σ 2
R=
 Squaring both equations and adding them: 2
σ +σ2 
C = 1 , 0
 σ1 + σ 2 
2
 σ1 − σ 2 
2
REMARK  2 
 σ −  + τ 2
=
  This expression
 2   2 
is valid for any
Eq. of a circle with value of .
center C and radius R .
Mohr’s Circle

52
Mohr’s Circle for a 2D State of Stress

 The locus of the points representative of the


state of stress on any of the planes passing
through a given point P is a circle.
(Mohr’s Circle)
 The inverse is also true:
 Given a point (σ ,τ ) in Mohr’s Circle, there is a plane passing through
P whose normal and tangential stresses are and τ , respectively.
 σ1 + σ 2  σ1 − σ 2
 σ −  Mohr's 2D problem space R=
 2  σ −a 2
cos ( 2 β ) =
=
 σ1 − σ 2  R
 
 2 
τ τ
sin ( 2 β )
= =
 σ1 − σ 2  R σ +σ2 
  C = 1 , 0
 2   2 

53
Construction of Mohr’s Circle
 Interactive applets and animations:
 by M. Bergdorf:
http://www.zfm.ethz.ch/meca/applets/mohr/Mohrcircle.htm

 from MIT OpenCourseware:


http://ocw.mit.edu/ans7870/3/3.11/tools/mohrscircleapplet.html

 from Virginia Tech:


http://web.njit.edu/~ala/keith/JAVA/Mohr.html

 From Pennsilvania State University:


http://www.esm.psu.edu/courses/emch13d/design/animation/animation.htm

54
Mohr’s Circle’s Properties
A. To obtain the point in Mohr’s Circle representative of the state of
stress on a plane which forms an angle β with the principal stress
direction σ 1:
1. Begin at the point on the circle (representative of the plane where σ 1
acts).
2. Rotate twice the angle in the sense σ 1 → σ β .
3. This point represents the shear and normal stresses at the desired plane
(representative of the stress state at the plane where acts).

3.

2. 1.

55
Mohr’s Circle’s Properties
B. The representative points of the state of stress on two
orthogonal planes are aligned with the centre of Mohr’s Circle:
π
 This is a consequence of property A as β=
2 β1 + .
2

56
Mohr’s Circle’s Properties
C. If the state of stress on two orthogonal planes is known, Mohr’s
Circle can be easily drawn:
1. Following property B, the two points representative of these planes will
be aligned with the centre of Mohr’s Circle.
2. Joining the points, the intersection with the σ axis will give the centre of
Mohr’s Circle.
3. Mohr’s Circle can be drawn.
3.
1.
2.

57
Mohr’s Circle’s Properties
D. Given the components of the stress tensor in a particular
orthonormal base, Mohr’s Circle can be easily drawn:
 This is a particular case of property C in which the points
representative of the state of stress on the Cartesian planes is known.
1. Following property B, the two points representative of these planes will
be aligned with the centre of Mohr’s Circle.
2. Joining the points, the intersection with the σ axis will give the centre of
Mohr’s Circle.
3. Mohr’s Circle can be drawn. 3.
1.
2.

σ x τ xy 
σ= 
τ xy σ y 
58
Mohr’s Circle’s Properties
 The radius and the diametric points of the circle can be obtained:

σ x τ xy 
σ= 
τ
 xy σ y

 σx −σ y 
2

=  + τ xy
2
R 
 2 

59
Mohr’s Circle’s Properties
 Note that the application of property A for the point
representative of the vertical plane implies rotating in
the sense contrary to angle.

σ x τ xy 
σ= 
τ xy σ y 

60
The Pole or the Origin of Planes
 The point called pole or origin of planes in Mohr’s circle has the
following characteristics:
 Any straight line drawn from the pole will intersect the Mohr circle at a
point that represents the state of stress on a plane parallel in space to
that line.
2.

1.

61
The Pole or the Origin of Planes
 The point called pole or origin of planes in Mohr’s circle has the
following characteristics:
 If a straight line, parallel to a given plane, is drawn from the pole, the
intersection point represents the state of stress on this particular plane.

1.

2.

62
Sign Convention in Soil Mechanics
 The sign criterion used in soil mechanics, is the inverse of the one
used in continuum mechanics:
 In soil mechanics,
negative (−) tensile stress
σβ
positive (+) compressive stress

positive (+) counterclockwise rotation continuum mechanics


τβ
negative (-) clockwise rotation

 But the sign criterion for angles is the same:


positive angles are measured counterclockwise
τ β* = −τ β
σ β* = −σ β
soil mechanics

63
Sign Convention in Soil Mechanics
 For the same stress state, the principal stresses will be inverted.

σ 1* = −σ 2
τ β* = −τ β σ 2* = −σ 1
σ β* = −σ β β *= β +
π
continuum mechanics 2
soil mechanics

 The expressions for the normal and shear stresses are


 * −σ 2* − σ 1* −σ 2* + σ 1* 
=−
 βσ + cos

(

2 β

*
+ π

) 
σ1 + σ 2 σ1 − σ 2   2 2  * σ 1* + σ 2* σ 1* − σ 2*
σβ = + cos ( 2 β )  −  2β * 

σβ = + cos ( 2β * )
2 2   cos 


   2 2
→ →
σ −σ  −τ * −σ 2 + σ 1 sin 2β * + π
* *
 τ * σ 1 − σ 2 sin 2β *
* *
τ β = 1 2 sin ( 2β )
2
=
  β
2 
(
 
) =
  β 2 
( )
  
− sin  2 β *  like in
   
continuum mechanics
 The Mohr’s circle construction and properties are the same in both
cases
64
4.10. Particular Cases of Mohr’s Circle
Ch.4. Stress

65
Particular Cases of Mohr’s Circles
 Hydrostatic state of stress

 Mohr’s circles of a stress tensor and its deviator


(σ sph = σ m 1) σ=
1 σ m + σ 1′
σ σ sph + σ′
= σ=2 σ m + σ 2′
σ=
3 σ m + σ 3′

 Pure shear state of stress

66
Chapter 4
Stress

rs
ee
s gin
4.1 Forces Acting on a Continuum Body

t d le En

r
Two types of forces that can act on a continuous medium will be considered:

ba
ge ro or
eS m
body forces and surface forces.

ci
f

ra
C d P cs
4.1.1 Body Forces
b
a
i
an an n
y ha

Definition 4.1. The body forces are the forces that act at a distance
le
liv or ec

on the internal particles of a continuous medium. Examples of this


M

.A

kind of forces are the gravitational, inertial or magnetic attraction


forces.
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 4.1: Body forces on a continuous medium.

127
128 C HAPTER 4. S TRESS

Consider b (x,t) is the spatial description of the vector field of body forces
per unit of mass. Multiplying the vector of body forces b (x,t) by the density ρ,
the vector of body forces per unit of volume ρb (x,t) (density of body forces) is
obtained. The total resultant, fV , of the body forces on the material volume V in
Figure 4.1 is 
fV = ρb (x,t) dV . (4.1)
V

Remark 4.1. In the definition of body forces given in (4.1), the exis-

rs
tence of the vector density of body forces ρb (x,t) is implicitly ac-

ee
cepted. This means that, given an arbitrary sequence of volumes ΔVi

s gin
that contain the particle P, and the corresponding sequence of body
forces fΔVi , there exists the limit

t d le En
fΔVi
ρb (x,t) = lim
ΔVi →0 ΔVi

r
ba
ge ro or
and, in addition, it is independent of the sequence of volumes con-

eS m
ci
sidered.
f

ra
C d P cs
b
a
i
an an n
y ha

Example 4.1 – Given a continuous medium with volume V placed on the


le
liv or ec

Earth’s surface, obtain the value of the total resultant of the body forces in
terms of the gravitational constant g.
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Solution
Assuming a system of Cartesian axes (see figure above) such that the x3 -
axis is in the direction of the vertical from the center of the Earth, the vector
field b (x,t) of gravitational force per unit of mass is
b (x,t) ≡ [ 0 , 0 , −g ]T
not

and, finally, the vector of body forces is 


  T
not
fV = ρb (x,t) dV ≡ 0 , 0 , − ρg dV .
V
V

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Forces Acting on a Continuum Body 129

4.1.2 Surface Forces

Definition 4.2. The surface forces are the forces that act on the
boundary of the material volume considered. They can be regarded
as produced by the contact actions of the particles located in the
boundary of the medium with the exterior of this medium.

Consider the spatial description of the vector field of surface forces per unit of

rs
surface t (x,t) on the continuous medium shown in Figure 4.2. The resultant

ee
force on a differential surface element dS is t dS and the total resultant of the

s gin
surface forces acting on the boundary ∂V of volume V can be written as

fS = t (x,t) dS .

t d le En
(4.2)
∂V

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
Remark 4.2. In the definition of surface forces given in (4.2), the ex-
b
a
istence of the vector of surface forces per unit of surface t (x,t) (trac-
i
an an n

tion vector1 ) is implicitly accepted. In other words, if a sequence of


y ha

surfaces Δ Si , each containing point P, and the corresponding surface


le
liv or ec

forces fΔ Si are considered (see Figure 4.3), there exists the limit
M

.A

fΔ S i
t (x,t) = lim
Δ Si →0 Δ Si
m

and it is independent of the chosen sequence of surfaces.


uu
e
X Th

er
tin
on

.O
C

Figure 4.2: Surface forces on a continuous medium.

1 In literature, the vector of surface forces per unit of surface t (x,t) is often termed traction
vector, although this concept can be extended to points in the interior of the continuous
medium.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
130 C HAPTER 4. S TRESS

Figure 4.3: Traction vector.

rs
ee
4.2 Cauchy’s Postulates

s gin
Consider a continuous medium on which body and surface forces are acting (see

t d le En
Figure 4.4). Consider also a particle P in the interior of the continuous medium

r
and an arbitrary surface containing point P and with a unit normal vector n at

ba
ge ro or
eS m
this point, which divides the continuous medium into two parts (material vol-

ci
f
umes). The surface forces due to the contact between volumes will act on the

ra
C d P cs
imaginary separating surface, considered now a part of the boundary of each of
b
a
i
these material volumes.
an an n

Consider the traction vector t that acts at the chosen point P as part of the
y ha

boundary of the first material volume. In principle, this traction vector (de-
le
liv or ec

fined now at a material point belonging to the interior of the original continuous
M

.A

medium) will depend on


1) the particle being considered,
m

2) the orientation of the surface (defined by means of the normal n) and


d
uu

3) the separating surface itself.


e
X Th

er
tin
on

.O
C

Figure 4.4: Cauchy’s postulates.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Cauchy’s Postulates 131

The following postulate2 makes it independent of this last condition.

Definition 4.3. Cauchy’s 1st postulate establishes that the traction


vector that acts at a material point P of a continuous medium ac-
cording to a plane with unit normal vector n depends only on the
point P and the normal n.

t = t (P, n)

rs
ee
s gin
t d le En

r
ba
ge ro or
Remark 4.3. Consider a particle P of a continuous medium and dif-

eS m
ci
ferent surfaces that contain this point P such that they all have the
f

ra
same unit normal vector n at said point. In accordance with Cauchy’s
C d P cs
b
a
postulate, the traction vectors at point P, according to each of these
i
an an n

surfaces, coincide. On the contrary, if the normal to the surfaces at P


y ha

is different, the corresponding traction vectors will not coincide (see


le
Figure 4.5).
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 4.5: Traction vector at a point according to different surfaces.

2 A postulate is a fundamental ingredient of a theory that is formulated as a principle of this


theory and, as such, does not need proof.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
132 C HAPTER 4. S TRESS

Definition 4.4. Cauchy’s 2nd postulate - action and reaction law es-
tablishes the traction vector at point P of a continuous medium, ac-
cording to a plane with unit normal vector n, has the same magnitude
and opposite direction to the traction vector at the same point P ac-
cording to a plane with unit normal vector −n at the same point (see
Figure 4.4).
t (P, n) = −t (P, −n)

rs
ee
s gin
4.3 Stress Tensor

t d le En
4.3.1 Application of Newton’s 2nd Law to a Continuous Medium

r
ba
ge ro or
Consider a discrete system of particles in motion such that a generic particle i

eS m
of this system has mass mi , velocity vi and acceleration ai = dvi /dt. In addition,

ci
f

ra
a force fi acts on each particle i, which is related to the particle’s acceleration
C d P cs
b
a
through Newton’s second law3 ,
i
an an n
y ha

fi = mi ai . (4.3)
le
liv or ec

Then, the resultant R of the forces that act on all the particles of the system is
M

.A

R = ∑ fi = ∑ mi ai . (4.4)
m

i i
d
uu
e

The previous concepts can be generalized for the case of continuous mediums
X Th

er
tin

when these are understood as discrete systems constituted by an infinite number


of particles. In this case, the application of Newton’s second law to a continu-
on

.O

ous medium with total mass M, on which external forces characterized by the
vector density of body forces ρb (x,t) and the traction vector t (x,t) are acting,
C

whose particles have an acceleration a (x,t), and that occupies at time t the space
©

volume Vt results in
   
R= ρb dV + t dS = a dm = ρa dV . (4.5)
 V
Vt ∂Vt M
    ρdV t

Resultant of Resultant of
the body the surface
forces forces

3 The Einstein notation introduced in (1.1) is not used here.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Stress Tensor 133

4.3.2 Stress Tensor


Consider now the particular case of a material volume constituted by an ele-
mental tetrahedron placed in the neighborhood of an arbitrary particle P of the
interior of the continuous medium and oriented according to the scheme in Fig-
ure 4.6. Without loss of generality, the origin of coordinates can be placed at P.
The tetrahedron has a vertex at P and its faces are completely defined by
means of a plane with normal n = [n1 , n2 , n3 ]T that intersects with the coordi-
nate planes, defining a generic surface with area S (the base of the tetrahedron)
at a distance h (the height of the tetrahedron) of point P. In turn, the coordinate
planes define the other faces of the tetrahedron with areas S1 , S2 and S3 , and

rs
(outward) normals −ê1 , −ê2 and −ê3 , respectively. Through geometric consid-

ee
erations, the relations

s gin
S1 = n1 S S2 = n2 S S3 = n3 S (4.6)

t d le En
can be established. The notation for the traction vectors on each of the faces of

r
the tetrahedron is introduced in Figure 4.7 as well as the corresponding normals

ba
ge ro or
eS m
with which they are associated.

ci
f

ra
According to Cauchy’s second postulate (see Definition 4.4), the traction vec-
C d P cs
b
a
tor on a generic point x belonging to one of the surfaces Si (with outward nor-
i
mal −êi ) can be written as
an an n
y ha

t (x, −êi ) = −t (x, êi ) = −t(i) (x)


not
i ∈ {1, 2, 3} .
le
(4.7)
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 4.6: Elemental tetrahedron in the neighborhood of a material point P.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
134 C HAPTER 4. S TRESS

rs
ee
s gin
Figure 4.7: Traction vectors on an elemental tetrahedron.

t d le En

r
ba
ge ro or
eS m
ci
f
Remark 4.4. The mean value theorem establishes that, given a

ra
C d P cs
(scalar, vectorial o tensorial) function that is continuous in the in-
b
a
i
terior of a (compact) domain, the function reaches its mean value
an an n

in the interior of said domain. In mathematical terms, given f (x)


y ha

continuous in Ω ,
le
liv or ec


∃ x∗ ∈ Ω | f (x) dΩ = Ω · f (x∗ )
M

.A

Ω
m

where f (x∗ ) is the mean value of f in Ω . Figure 4.8 shows the


uu
e
X Th

graphical interpretation of the mean value theorem in one dimen-


er
tin

sion.
on

.O
C

Figure 4.8: Mean value theorem.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Stress Tensor 135

In virtue of the mean value theorem, the vector field t(i) (x), assumed to be
continuous in the domain Si , attains its mean value in the interior of this domain.



Let x∗sI ∈ Si be the point where the mean value is reached and t(i) = t(i) x∗sI


this mean value. Analogously, the vectors t∗ = t x∗S , ρ ∗ b∗ = ρ (xV∗ ) b (xV∗ )
and ρ ∗ a∗ = ρ (xV∗ ) a (xV∗ ) are the mean values corresponding to the vector fields:
traction vector t (x) in S, density of body forces ρb (x) and inertial forces ρa (x),
respectively. These mean values are attained, again according to the mean value
theorem, at points x∗s ∈ S and xV∗ ∈ V of the interior of the corresponding do-
mains. Therefore, one can write

rs
 

t(i) (x) dS = t(i) Si i ∈ {1, 2, 3} , t (x) dS = t∗ S ,

ee
s gin
Si S
(4.8)
 
ρ (x) b (x) dV = ρ ∗ b∗V ρ (x) a (x) dV = ρ ∗ a∗V .

t d le En
and

r
V V

ba
ge ro or
eS m
ci
Applying now (4.5) on the tetrahedron considered, results in
f

ra
    
C d P cs
b
a
ρb dV + t dS + t dS + t dS + t dS =
i
an an n
y ha

V S S1 S2 S3
      (4.9)
le
= ρb dV + t dS + −t(1) dS + −t(2) dS + −t(3) dS = ρa dV,
liv or ec
M

.A

V S S1 S2 S3 V
m

where (4.7) has been taken into account. Replacing (4.8) in (4.9), the latter can
d

be written in terms of the mean values as


uu
e
X Th

∗ ∗ ∗
ρ ∗ b∗ V + t∗ S − t(1) S1 − t(2) S2 − t(3) S3 = ρ ∗ a∗ V .
er
tin

(4.10)
on

.O

Introducing now (4.6) and expressing the total volume of the tetrahedron as
V = Sh/3, the equation above becomes
C

1 ∗ ∗ ∗ ∗ ∗ 1
ρ b h S + t∗ S − t(1) n1 S − t(2) n2 S − t(3) n3 S = ρ ∗ a∗ h S =⇒
3 3 (4.11)
1 ∗ ∗ ∗ (1)∗
(2) ∗
(3) ∗ 1 ∗ ∗
ρ b h + t − t n1 − t n2 − t n3 = ρ a h .
3 3
Expression (4.11) is valid for any tetrahedron defined by a plane with unit
normal vector n placed at a distance h of point P. Consider now an infinites-
imal
tetrahedron,
also in the neighborhood of point P, by making the value
of PP = h tend to zero but maintaining the orientation of the plane constant
(n=constant). Then, the domains Si , S and V in (4.11) collapse into point P (see
Figure 4.7). Therefore, the points of the corresponding domains in which the

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
136 C HAPTER 4. S TRESS

mean values are obtained also tend to point P,




x∗Si → xP =⇒ lim t(i) x∗Si = t(i) (P) i ∈ {1, 2, 3} ,
h→0
(4.12)
x∗S → xP =⇒ lim t∗ (x∗S , n) = t (P, n) ,
h→0

and, in addition,
 
1 ∗ ∗ 1 ∗ ∗
lim ρ b h = lim ρ a h =0. (4.13)
h→0 3 h→0 3

rs
ee
Taking the limit of (4.11) and replacing expressions (4.12) and (4.13) in it
leads to

s gin
t (P, n) − t(1) n1 − t(2) n2 − t(3) n3 = 0 =⇒ t (P, n) − t(i) ni = 0 . (4.14)

t d le En
The traction vector t(1) can be written in terms of its corresponding Cartesian

r
ba
ge ro or
eS m
components (see Figure 4.9) as

ci
f

ra
C d P cs
t(1) = σ11 ê1 + σ12 ê2 + σ13 ê3 = σ1i êi . (4.15)
b
a
i
an an n

Operating in an analogous manner on traction vectors t(2) and t(3) (see Fig-
y ha

ure 4.10) results in


le
liv or ec

t(2) = σ21 ê1 + σ22 ê2 + σ23 ê3 = σ2i êi (4.16)
M

.A

t(3) = σ31 ê1 + σ32 ê2 + σ33 ê3 = σ3i êi


m

(4.17)
d
uu
e

and, for the general case,


X Th

er
tin

t(i) (P) = σi j ê j i, j ∈ {1, 2, 3} . (4.18)


on

.O

(i)
σi j (P) = t j (P) i, j ∈ {1, 2, 3}
C

(4.19)
©

Remark 4.5. Note that in expression (4.19) the functions σi j are


(i)
functions of (the components of) the traction vectors t j (P) on the
surfaces specifically oriented at point P. Thus, it is emphasized that
these functions depend on point P but not on the unit normal vec-
tor n.
σi j = σi j (P)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Stress Tensor 137

Figure 4.9: Decomposition of the traction vector t(1) into its components.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n

Figure 4.10: Traction vectors t(2) and t(3) .


y ha

le
liv or ec
M

.A

Replacing (4.19) in (4.14) yields


(i)
m

t (P, n) = ni t(i) =⇒ t j (P, n) = ni t j (P) = ni σi j (P) i, j ∈ {1, 2, 3} =⇒


d
uu
e
X Th

t (P, n) = n · σ (P)
er

(4.20)
tin
on

where the Cauchy stress tensor σ is defined as


.O
C

σ = σi j êi ⊗ ê j . (4.21)
©

Remark 4.6. Note that expression (4.20) is consistent with Cauchy’s


first postulate (see Definition 4.3) and that the second postulate (see
Definition 4.4) is satisfied from

t (P, n) = n · σ ⎬
=⇒ t (P, n) = −t (P, −n) .
t (P, −n) = −n · σ ⎭

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
138 C HAPTER 4. S TRESS

Figure 4.11: Traction vectors for the construction of the Cauchy stress tensor.

rs
ee
s gin
Remark 4.7. In accordance with (4.18) and (4.21), the Cauchy stress

t d le En
tensor is constructed from the traction vectors according to three co-
ordinate planes that include point P (see Figure 4.11). However, by

r
ba
means of (4.20), the stress tensor σ (P) is seen to contain informa-

ge ro or
eS m
ci
tion on the traction vectors corresponding to any plane (identified by
f

ra
its normal n) that contains this point.
C d P cs
b
a
i
an an n
y ha

le
liv or ec

4.3.3 Graphical Representation of the Stress State in a Point


M

.A

It is common to resort to graphical representations of the stress tensor based on


m

elemental parallelepipeds in the neighborhood of the particle considered, with


d

faces oriented in accordance to the Cartesian planes and in which the corre-
uu
e

sponding traction vectors are decomposed into their normal and tangent compo-
X Th

er
tin

nents following expressions (4.15) through (4.20) (see Figure 4.12).


on

.O

4.3.3.1 Scientific Notation


C

The representation in Figure 4.12 corresponds to what is known as scientific


©

notation. In this notation, the matrix of components of the stress tensor is written
as ⎡ ⎤
σ11 σ12 σ13
not ⎢ ⎥
σ ≡ ⎣ σ21 σ22 σ23 ⎦ (4.22)
σ31 σ32 σ33
and each component σi j can be characterized in terms of its indices:
− Index i indicates the plane on which the stress acts (plane perpendicular to
the xi -axis).
− Index j indicates the direction of the stress (direction of the x j -axis).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Stress Tensor 139

rs
ee
Figure 4.12: Graphical representation of the stress tensor (scientific notation).

s gin
t d le En
4.3.3.2 Engineering Notation
In engineering notation, the components of the Cauchy stress tensor (see Fig-

r
ba
ge ro or
ure 4.13) are written as

eS m
⎡ ⎤

ci
σx τxy τxz
f

ra
not ⎢ ⎥
C d P cs
σ ≡ ⎣ τyx σy τyz ⎦
b
a
(4.23)
i
an an n

τzx τyz σz
y ha

le
and each component can be characterized as follows:
liv or ec

− The component σa is the normal stress acting on the plane perpendicular to


M

.A

the a-axis.
− The component τab is the tangential (shear) stress acting on the plane per-
m

pendicular to the a-axis in the direction of the b-axis.


uu
e
X Th

er
tin
on

.O
C

Figure 4.13: Graphical representation of the stress tensor (engineering notation).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
140 C HAPTER 4. S TRESS

4.3.3.3 Sign Criterion


Consider a particle P of the continuous medium and a plane with unit normal
vector n that contains this particle (see Figure 4.14). The corresponding traction
vector t can be decomposed into its normal component σ n and its tangential
component τ n . The sign of the projection of t on n (σ = t · n) defines the tensile
σ n tends to pull on the plane ) or compressive (σ
(σ σ n tends to compress the plane)
character of the normal component.
This concept can be used to define the sign of the components of the stress
tensor. For this purpose, in the elemental parallelepiped of Figure 4.12, the dis-
tinction is made between the positive or visible faces (its outward normal has

rs
the same direction as the positive base vector and the faces can be seen in the

ee
figure) and the negative or hidden faces.
The sign criterion for the visible faces is

s gin

positive (+) ⇒ tension

t d le En
Normal stresses σi j or σa and
negative (−) ⇒ compression

r
ba
ge ro or
eS m


ci
positive (+) ⇒ direction of b-axis
Tangential stresses τab f

ra
C d P cs
negative (−) ⇒ opposite direction to b-axis
b
a
i
an an n

In accordance with this criterion, the directions of the stresses represented in


y ha

Figure 4.13 (on the visible faces of the parallelepiped) correspond to positive
le
liv or ec

values of the respective components of the stress tensor4 .


In virtue of the action and reaction law (see Definition 4.4) and for the hidden
M

.A

faces of the parallelepiped, the aforementioned positive values of the compo-


m

nents of the stress tensor correspond to opposite directions in their graphical


d

representation (see Figure 4.15).


uu
e
X Th

er
tin
on

.O
C

σ n = σn

> 0 tension
σ = t·n
< 0 compression

Figure 4.14: Decomposition of the traction vector.

4 It is obvious that the negative values of the components of the stress tensor will result in
graphical representations of opposite direction to the positive values indicated in the figures.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Properties of the Stress Tensor 141

Figure 4.15: Positive stresses in the hidden faces.

rs
ee
4.4 Properties of the Stress Tensor

s gin
Consider an arbitrary material volume V in a continuous medium and its bound-

t d le En
ary ∂V . The body forces b (x,t) act on V and the prescribed traction vector
t∗ (x,t) acts on ∂V . The acceleration vector field of the particles is a (x,t) and

r
ba
ge ro or
eS m
the Cauchy stress tensor field is σ (x,t) (see Figure 4.16).

ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O

Figure 4.16: Forces acting on a continuous medium.


C

4.4.1 Cauchy Equation. Internal Equilibrium Equation


The stress tensor, the body forces and the accelerations are related through
Cauchy’s equation,


⎨ ∇ · σ + ρb = ρa ∀x ∈ V
Cauchy’s
equation ⎪ ∂ σi j (4.24)
⎩ + ρb j = ρa j j ∈ {1, 2, 3}
∂ xi

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
142 C HAPTER 4. S TRESS

whose explicit expression in engineering notation is




⎪ ∂ σx ∂ τyx ∂ τzx

⎪ + + + ρbx = ρax ,

⎪ ∂x ∂y ∂z


∂ τxy ∂ σy ∂ τzy
⎪ + + + ρby = ρay , (4.25)

⎪ ∂x ∂y ∂z



⎩ ∂ τxz + ∂ τyz + ∂ σz + ρbz = ρaz .

∂x ∂y ∂z

rs
If the system is in equilibrium, the acceleration is null (a = 0), and (4.24) is

ee
reduced to

s gin
Internal ⎨ ∇ · σ + ρb = 0 ∀x ∈ V
∂ σi j

t d le En
equilibrium (4.26)
equation ⎩ ∂ x + ρb j = 0 j ∈ {1, 2, 3}

r
i

ba
ge ro or
eS m
ci
f
which is known as the internal equilibrium equation of the continuous medium.

ra
C d P cs
Cauchy’s equation of motion is derived from the principle of balance of linear
b
a
i
momentum, which will be studied in Chapter 5.
an an n
y ha

le
4.4.2 Equilibrium Equation at the Boundary
liv or ec

Equation (4.20) is applied on the boundary points taking into account that the
M

.A

traction vector is now known in said points (t = t∗ ). The result is denoted as


m

equilibrium equation at the boundary.


d
uu


e

Equilibrium n (x,t) · σ (x,t) = t∗ (x,t)


X Th

∀x ∈ ∂V
er
tin

equation at (4.27)
the boundary ni σi j = t ∗j j ∈ {1, 2, 3}
on

.O
C

4.4.3 Symmetry of the Cauchy Stress Tensor


The Cauchy stress tensor is proven to be symmetric by applying the principle of
balance of angular momentum (see Chapter 5).

σ = σT
(4.28)
σi j = σ ji i, j ∈ {1, 2, 3}

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Properties of the Stress Tensor 143

Remark 4.8. The symmetry of the stress tensor allows the Cauchy’s
equation (4.24) and the equilibrium equation at the boundary (4.27)
to be written, respectively, as

⎨ ∇ · σ + ρb = σ · ∇ + ρb = ρa ∀x ∈ V
∂σ ∂ σ ji
⎩ i j + ρb j = + ρb j = ρa j j ∈ {1, 2, 3}
∂ xi ∂ xi

n · σ = σ · n = t∗ (x,t) ∀x ∈ ∂V

rs
ni σi j = σ ji ni = t j∗ j ∈ {1, 2, 3}

ee
s gin
t d le En

r
Example 4.2 – A continuous medium moves with a velocity field whose spa-

ba
ge ro or
eS m
tial description is v (x,t) ≡ [z, x, y]T . The Cauchy stress tensor is
not

ci
f

ra
⎡ ⎤
C d P cs
y g (x, z,t) 0
b
a
not ⎢ ⎥
i
an an n

σ ≡ ⎣ h (y) z (1 + t) 0 ⎦ .
y ha

0 0 0
le
liv or ec

Determine the functions g, h and the spatial form of the body forces b (x,t)
M

.A

that generate the motion.


m

d
uu

Solution
e
X Th

er
tin

The stress tensor is symmetric, therefore



on

.O

h (y) = C ,
σ =σ T
=⇒ h (y) = g (x, z,t) =⇒
g (x, z,t) = C ,
C

where C is a constant. In addition, the divergence of the tensor is null,


⎡ ⎤
  y C 0
not ∂ ∂ ∂ ⎢ ⎥
∇·σ ≡ , , ⎣ C z (1 + t) 0 ⎦ = [0, 0, 0] .
∂x ∂y ∂z
0 0 0

Thus, Cauchy’s equation is reduced to

∇ · σ + ρb = ρa
=⇒ b=a.
∇·σ = 0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
144 C HAPTER 4. S TRESS

Applying the expression for the material derivative of velocity,


dv ∂ v
a= = + v · ∇v with
dt ∂t
⎡ ⎤

⎢ ⎥ ⎡ ⎤
⎢ ∂x ⎥
∂v ⎢ ⎥ 0 1 0
not ⎢ ⎥
=0 and ∇v = ∇ ⊗ v ≡ ⎢ ∂ ⎥ [z, x, y] = ⎣ 0 0 1 ⎦ .
∂t ⎢ ∂y ⎥
⎢ ⎥ 1 0 0
⎣ ∂ ⎦

rs
∂z

ee
s gin
the acceleration
⎡ ⎤
0 1 0

t d le En
a = v · ∇v ≡ [z, x, y] ⎣ 0 0 1 ⎦ = [y, z, x]
not

r
ba
ge ro or
1 0 0

eS m
ci
f

ra
is obtained. Finally, the body forces are
C d P cs
b
a
i
an an n

b (x,t) = a (x,t) ≡ [y, z, x]T .


not
y ha

le
liv or ec
M

.A

4.4.4 Diagonalization. Principal Stresses and Directions


m

Consider the stress tensor σ . Since it is a symmetric second-order tensor, it


uu
e

diagonalizes5 in an orthonormal basis and its eigenvalues are real. Consider,


X Th

er
tin

then, its matrix of components in the Cartesian basis {x, y, z} (see Figure 4.17),
⎡ ⎤
on

.O

σx τxy τxz
not ⎢ ⎥
C

σ ≡ ⎣ τyx σy τyz ⎦ . (4.29)


©

τzx τyz σz {x, y, z}

In the Cartesian system {x , y , z } in which σ diagonalizes, its matrix of com-


ponents will be
⎡ ⎤
σ1 0 0
not ⎢ ⎥
σ ≡ ⎣ 0 σ2 0 ⎦ . (4.30)
0 0 σ3 {x , y , z }
5 A theorem of tensor algebra guarantees that all symmetric second-order tensor diagonalizes
in an orthonormal basis and its eigenvalues are real.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Properties of the Stress Tensor 145

rs
ee
Figure 4.17: Diagonalization of the stress tensor.

s gin
t d le En

r
Definition 4.5. The principal stress directions are the directions, as-

ba
ge ro or
sociated with the axes {x , y , z }, in which the stress tensor diago-

eS m
ci
nalizes.
f

ra
C d P cs
The principal stresses are the eigenvalues of the stress tensor
b
a
(σ1 , σ2 , σ3 ). In general, they will be assumed to be arranged in the
i
an an n

form σ1 ≥ σ2 ≥ σ3 .
y ha

le
liv or ec
M

.A

To obtain the principal stress directions and the principal stresses, the eigen-
value problem associated with tensor σ must be posed. That is, if λ and v are an
m

eigenvalue and its corresponding eigenvector, respectively, then


uu
e
X Th

σ ·v = λv =⇒ σ − λ 1) · v = 0 .

er

(4.31)
tin

The solution to this system will not be trivial (will be different to v = 0) when
on

.O

the determinant of (4.31) is equal to zero, that is


C

not
det (σ σ − λ 1| = 0 .
σ − λ 1) = |σ (4.32)
Equation (4.32) is a third-grade polynomial equation in λ . Since tensor σ
is symmetric, its three solutions (λ1 ≡ σ1 , λ2 ≡ σ2 , λ3 ≡ σ3 ) are real. Once the
eigenvalues have been found and ordered according to the criterion σ1 ≥ σ2 ≥ σ3 ,
the eigenvector v(i) can be obtained for each stress σi by resolving the system in
(4.31),
σ − σi 1) · v(i) = 0
(σ i ∈ {1, 2, 3} . (4.33)
This equation provides a non-trivial solution of the eigenvectors v(i) , orthogo-
nal between themselves, which, once it has been normalized, defines the three
elements of the base corresponding to the three principal directions.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
146 C HAPTER 4. S TRESS

Remark 4.9. In accordance with the graphical interpretation of the


components of the stress tensor in Section 4.3.3, only normal stresses
act on the faces of the elemental parallelepiped associated with the
principal stress directions, which are, precisely, the principal stresses
(see Figure 4.17).

rs
4.4.5 Mean Stress and Mean Pressure

ee
s gin
Definition 4.6. The mean stress is the mean value of the principal

t d le En
stresses.

r
1

ba
σm = (σ1 + σ2 + σ3 )

ge ro or
eS m
3

ci
f

ra
C d P cs
b
a
i
an an n

Considering the matrix of components of the stress tensor in the principal stress
y ha

directions (4.30), results in


le
liv or ec

1 1
σm = σ) .
(σ1 + σ2 + σ3 ) = Tr (σ (4.34)
M

.A

3 3
m

d
uu
e

Definition 4.7. The mean pressure is the mean stress with its sign
X Th

er
tin

changed.
not 1
mean pressure = p̄ = −σm = − (σ1 + σ2 + σ3 )
on

3
.O
C

Definition 4.8. A spherical or hydrostatic stress state is a state in


which all three principal stress directions have the same value.
⎡ ⎤
σ 0 0
σ1 = σ2 = σ3 =⇒ σ ≡ ⎣ 0 σ 0 ⎦ ≡ σ 1
not not

0 0 σ

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Properties of the Stress Tensor 147

Remark 4.10. In a hydrostatic stress state, the stress tensor is


isotropic6 and, thus, its components are the same in every Cartesian
coordinate system.
As a consequence, any direction is a principal stress direction and
the stress state (traction vector) is the same in any plane.

rs
4.4.6 Decomposition of the Stress Tensor into its Spherical and

ee
Deviatoric Parts

s gin
The stress tensor σ can be split7 into a spherical part (or component) σ sph and
a deviatoric part σ  ,

t d le En
σ = σ sph + σ  . (4.35)
 

r
ba
ge ro or
eS m
spherical deviatoric

ci
part part
f

ra
C d P cs
b
a
The spherical part is defined as
i
an an n

⎡ ⎤
σm 0
y ha

0
de f 1 not ⎢ ⎥
le
σ sph : = σ ) 1 = σ m 1 ≡ ⎣ 0 σm 0 ⎦ ,
Tr (σ (4.36)
liv or ec

3
0 0 σm
M

.A
m

where σm is the mean stress defined in (4.34). According to definition (4.35), the
d
uu

deviatoric part of the stress tensor is


e
X Th

⎡ ⎤ ⎡ ⎤
er
tin

σx τxy τxz σm 0 0
not ⎢ ⎥ ⎢ ⎥
σ  = σ − σ sph ≡ ⎣ τxy σy τyz ⎦ − ⎣ 0 σm 0 ⎦
on

.O

(4.37)
τxz τyz σz 0 σm
C

0
©

resulting in
⎡ ⎤ ⎡ ⎤
σx − σm τxy τxz σx  τxy  τxz 
not ⎢ ⎥ ⎢  ⎥
σ ≡ ⎣ τxy σy − σm τyz ⎦ = ⎣ τxy σy  τyz  ⎦ . (4.38)
τxz τyz σz − σm τxz  τyz  σz 

6 A tensor is defined as isotropic when it remains invariant under any change of orthogonal
basis. The general expression of an isotropic second-order tensor is T = α1 where α can be
any scalar.
7 This type of decomposition can be applied to any second-order tensor.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
148 C HAPTER 4. S TRESS

Remark 4.11. The spherical part of the stress tensor σ sph is an


isotropic tensor (and defines a hydrostatic stress state), therefore, it
remains invariant under any change of orthogonal basis.

Remark 4.12. The deviatoric component of the tensor is an indica-


tor of how far from a hydrostatic stress state the present state is

rs
(see (4.37) and Remark 4.11).

ee
s gin
t d le En

r
Remark 4.13. The principal directions of the stress tensor and of its

ba
ge ro or
eS m
deviatoric tensor coincide. Proof is trivial considering that, from Re-

ci
mark 4.11, the spherical part σ sph is diagonal in any coordinate sys-
f

ra
C d P cs
tem. Consequently, if σ diagonalizes for a certain basis in (4.37), σ 
b
a
i
will also diagonalize for that basis.
an an n
y ha

le
liv or ec
M

.A

Remark 4.14. The trace of the deviatoric (component) tensor is null.


m

Taking into account (4.34) and (4.37),


d
uu

Tr σ  = Tr σ − σ sph = Tr (σ σ ) − Tr σ sph = 3σm − 3σm = 0 .


X Th

er
tin
on

.O
C

4.4.7 Tensor Invariants


The three fundamental invariants of the stress tensor8 (or I invariants) are
σ ) = σii = σ1 + σ2 + σ3 ,
I1 = Tr (σ (4.39)
1

I2 = σ : σ − I12 = − (σ1 σ2 + σ1 σ3 + σ2 σ3 ) , (4.40)
2
σ) .
I3 = det (σ (4.41)

8 The tensor invariants are scalar algebraic combinations of the components of a tensor that
do not vary when the basis changes.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Stress Tensor in Curvilinear Orthogonal Coordinates 149

Any combination of the I invariants is, in turn, another invariant. In this manner,
the J invariants
J1 = I1 = σii , (4.42)
1
2 1 1
J2 = I1 + 2I2 = σi j σ ji = (σ σ : σ) , (4.43)
2 2 2
1
3 1 1
J3 = I1 + 3I1 I2 + 3I3 = Tr (σ σ · σ · σ ) = σi j σ jk σki , (4.44)
3 3 3
are defined.

rs
ee
Remark 4.15. For a purely deviatoric tensor σ  , the corresponding J

s gin
invariants are (see Remark 4.14 and equations (4.39) to (4.44))

t d le En
⎫ ⎪
⎪ J1  = I1  = 0


J1 = I1 = 0 ⎪
⎬ ⎪

r
ba
ge ro or
=⇒ σ  =⇒ J2  = I2  = 1 (σ 1

eS m
J2 = I2 σ  : σ  ) = σ  i j σ  ji

ci

⎭ ⎪

⎪ f2 2

ra
J3 = I3 ⎪
⎪ 

C d P cs
⎪ 1
b
⎩ J3 = I3  = σ  i j σ  jk σ  ki

a
i
3
an an n
y ha

le
liv or ec
M

.A

4.5 Stress Tensor in Curvilinear Orthogonal Coordinates


m

4.5.1 Cylindrical Coordinates


uu
e
X Th

Consider a point in space defined by the cylindrical coordinates {r, θ , z} (see


er
tin

Figure 4.18). A physical (orthonormal) basis {êr , êθ , êz } and a Cartesian system
of local axes {x , y , z } defined as dextrorotatory are considered at this point.
on

.O

The components of the stress tensor in this basis are


C

⎡ ⎤ ⎡ ⎤
©

σx τx y τx z σr τrθ τrz


not ⎢ ⎥ ⎢ ⎥
σ ≡ ⎣ τx y σy τy z ⎦ = ⎣ τrθ σθ τθ z ⎦ . (4.45)
τx z τy z σz τrz τθ z σz

The graphical representation on an elemental parallelepiped is shown in Fig-


ure 4.19, where the components of the stress tensor have been drawn on the
visible faces. Note that, here, the visible faces of the figure do not coincide with
the positive faces, defined (in the same direction as in Section 4.3.3.3) as those
whose unit normal vector has the same direction as a vector of the physical basis.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
150 C HAPTER 4. S TRESS

⎡ ⎤
x = r cos θ
x (r, θ , z) ≡ ⎣ y = r sin θ ⎦
not

z=z

rs
ee
Figure 4.18: Cylindrical coordinates.

s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

Figure 4.19: Differential element in cylindrical coordinates.


d
uu
e
X Th

er
tin

4.5.2 Spherical Coordinates


{r, θ , φ } (see Fig-
on

.O

A point in space is defined by the spherical ! coordinates


"
ure 4.20). A physical (orthonormal) basis êr , êθ , êφ and a Cartesian system
C

of local axes {x , y , z } defined as dextrorotatory are considered at this point.


©

The components of the stress tensor in this basis are


⎡ ⎤ ⎡ ⎤
σx τx y τx z σr τrθ τrφ
not ⎢ ⎥ ⎢ ⎥
σ ≡ ⎣ τx y σy τy z ⎦ = ⎣ τrθ σθ τθ φ ⎦ . (4.46)
τx z τy z σz τrφ τθ φ σφ

The graphical representation on an elemental parallelepiped is shown in Fig-


ure 4.21, where the components of the stress tensor have been drawn on the
visible faces.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Mohr’s Circle in 3 Dimensions 151

⎡ ⎤
x = r sin θ cos φ
x (r, θ , φ ) ≡ ⎣ y = r sin θ sin φ ⎦
not

z = z cos θ

rs
ee
s gin
Figure 4.20: Spherical coordinates.

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin

Figure 4.21: Differential element in spherical coordinates.


on

.O
C

4.6 Mohr’s Circle in 3 Dimensions


©

4.6.1 Graphical Interpretation of the Stress States


The stress tensor plays such a crucial role in engineering that, traditionally, sev-
eral procedures have been developed, essentially graphical ones, to visualize and
interpret it. The most common are the so-called Mohr’s circles.
Consider an arbitrary point in the continuous medium P and the stress tensor
σ (P) at this point. Consider also an arbitrary plane, with unit normal vector n,
that contains P (see Figure 4.22). The traction vector acting on point P corre-
sponding to this plane is t = σ · n. This vector can now be decomposed into its
components σ n , normal to the plane considered, and τ n , tangent to said plane.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
152 C HAPTER 4. S TRESS

Figure 4.22: Decomposition of the traction vector.

rs
ee
s gin
Consider now the normal component σ n = σ n, where σ is the normal com-
ponent of the stress on the plane, defined in accordance with the sign criterion

t d le En
detailed in Section 4.3.3.3,

σ > 0 tension ,

r
ba
σn = σ ·n

ge ro or
(4.47)

eS m
σ < 0 compression .

ci
f

ra
C d P cs
Consider now the tangential component τ n , of which only its module is of inter-
b
a
i
est,
an an n

τn = t−σn |ττ n | = τ ≥ 0 . (4.48)


y ha

The stress state on the plane with unit normal vector n at the point considered
le
liv or ec

can be characterized by means of the pair



M

.A

σ ∈R
(σ , τ) → (4.49)
τ ∈ R+
m

d
uu

which, in turn, determine a point of the half-plane (x ≡ σ , y ≡ τ) ∈ R × R+ in


e
X Th

er

Figure 4.23. If the infinite number of planes that contain point P are now con-
tin

sidered (characterized by all the possible unit normal vectors n(i) ) and the corre-
on

sponding values of the normal stress σi and tangential stress τi are obtained and,
.O

finally, are represented in the half-space mentioned above, a point cloud is ob-
C

tained. One can then wonder whether the point cloud occupies all the half-space
©

or is limited to a specific locus. The answer to this question is provided by the


following analysis.

n1 → (σ 1 , τ 1 )
n2 → (σ 2 , τ 2 )
· · ·
ni → (σ i , τ i )

Figure 4.23: Locus of points (σ , τ).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Mohr’s Circle in 3 Dimensions 153

4.6.2 Determination of the Mohr’s Circles


Consider the system of Cartesian axes associated with the principal directions
of the stress tensor. In this basis, the components of the stress tensor are
⎡ ⎤
σ1 0 0
not ⎢ ⎥
σ ≡ ⎣ 0 σ2 0 ⎦ with σ1 ≥ σ2 ≥ σ3 (4.50)
0 0 σ3

and the components of the traction vector are


⎡ ⎤⎡ ⎤ ⎡ ⎤

rs
σ1 0 0 n1 σ1 n1
not ⎢ ⎥⎢ ⎥ ⎢

ee

t = σ · n ≡ ⎣ 0 σ2 0 ⎦ ⎣ n2 ⎦ = ⎣ σ2 n2 ⎦ , (4.51)

s gin
0 0 σ3 n3 σ3 n3

t d le En
where n1 , n2 , n3 are the components of the unit normal vector n in the basis as-

r
sociated with the principal stress directions. In view of (4.51), the normal com-

ba
ge ro or
eS m
ponent of the stress (σ ), defined in (4.47), is

ci
⎡ ⎤ f

ra
C d P cs
n1
b
a
⎢ ⎥
i
not
t · n ≡ [σ1 n1 , σ2 n2 , σ3 n3 ] ⎣ n2 ⎦ = σ1 n21 + σ2 n22 + σ3 n23 = σ (4.52)
an an n
y ha

n3
le
liv or ec

and the module of the traction vector is


M

.A

|t|2 = t · t = σ12 n21 + σ22 n22 + σ32 n23 . (4.53)


m

The modules of the traction vector and of its normal and tangential components
uu
e

can also be related through


X Th

er
tin

|t|2 = σ12 n21 + σ22 n22 + σ32 n23 = σ 2 + τ 2 , (4.54)


on

.O

where (4.53) has been taken into account. Finally, the condition that n is a unit
C

normal vector can be expressed in terms of its components as


©

|n| = 1 =⇒ n21 + n22 + n23 = 1 . (4.55)

Equations (4.54), (4.52) and (4.55) can be summarized in the following ma-
trix equation.
⎡ 2 2 2 ⎤⎡ 2 ⎤ ⎡ 2 ⎤
σ1 σ2 σ3 n1 σ + τ2
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎣ σ1 σ2 σ3 ⎦ ⎣ n22 ⎦ = ⎣ σ ⎦ =⇒ A · x = b (4.56)
1 1 1 n23 1
     
A x b

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
154 C HAPTER 4. S TRESS

System (4.56) can be interpreted as a linear system with:


a) A matrix of coefficients, A (σσ ), defined by the stress tensor at point P
(by means of the principal stresses).
b) An independent term, b, defined by the coordinates of a certain point
in the half-space σ − τ (representative, in turn, of the stress state on a
certain plane).
c) A vector of unknowns x that determines (by means of the components
of the unit normal vector n) in which plane the values of the selected σ
and τ correspond.

rs
ee
Remark 4.16. Only the solutions of system (4.56) whose compo-

s gin
not # $T
nents x ≡ n21 , n22 , n23 are positive and smaller than 1 will be fea-

t d le En
sible (see (4.55)), i.e.,
0 ≤ n21 ≤ 1 , 0 ≤ n22 ≤ 1 0 ≤ n23 ≤ 1 .

r
and

ba
ge ro or
eS m
Every pair (σ , τ) that leads to a solution x that satisfies this require-

ci
f

ra
ment will be considered a feasible point of the half-space σ − τ,
C d P cs
b
a
which is representative of the stress state on a plane that contains P.
i
The locus of feasible points (σ , τ) is named feasible zone of the half-
an an n
y ha

space σ − τ.
le
liv or ec
M

.A

Consider now the goal of finding the feasible region. Through some algebraic
m

operations, system (4.56) can be rewritten as



d
uu

⎪ A
e


⎪ σ 2 + τ 2 − (σ1 + σ3 ) σ + σ1 σ3 − n2 = 0 (I)

X Th

⎪ (σ1 − σ3 ) 1
er


tin


A
σ 2 + τ 2 − (σ2 + σ3 ) σ + σ2 σ3 − n22 = 0 (II)
on

.O


⎪ (σ − σ )


2 3
(4.57)
⎪ A
C


⎩ σ 2 + τ 2 − (σ1 + σ2 ) σ + σ1 σ2 − n23 = 0 (III)
©

(σ1 − σ2 )

with A = (σ1 − σ2 ) (σ2 − σ3 ) (σ1 − σ3 ) .

Given, for example, equation (III) of the system in (4.57), it is easily verifiable
that it can be written as
1
(σ − a)2 + τ 2 = R2 with a = (σ1 + σ2 )
2
% (4.58)
1 2
and R = (σ1 − σ2 ) + (σ2 − σ3 ) (σ1 − σ3 ) n23 ,
4

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Mohr’s Circle in 3 Dimensions 155

which corresponds to the equation of a semicircle in the half-space σ − τ of


center C3 and a radius R3 , given by

1
C3 = (σ1 + σ2 ) , 0 and
2 (4.59)
%
1 2
R3 = (σ1 − σ2 ) + (σ2 − σ3 ) (σ1 − σ3 ) n23 .
4

The different values of n23 ∈ [0, 1] determine a set of concentric semicircles


of center C3 and radii R3 (n3 ) belonging to the half-space σ − τ and whose

rs
points occupy a certain region of this half-space. This region is delimited by

ee
the maximum and minimum values of R3 (n3 ). Observing that the radical in the

s gin
expression of R3 in (4.59) is positive, these values are obtained for n23 = 0 (the
minimum radius) and n23 = 1 (the maximum radius).

t d le En
1

r
n23 = 0 =⇒ R3min = (σ1 − σ2 )

ba
ge ro or
2

eS m
(4.60)

ci
1f

ra
n23 =1 =⇒ R3max = (σ1 + σ2 ) − σ3
C d P cs
b
a
2
i
an an n

The domain delimited by both semicircles defines an initial limitation of the


y ha

feasible domain, shown in Figure 4.24.


le
liv or ec

This process is repeated for the other two equations, (I) and (II), in (4.57),
M

.A

resulting in: ⎧
&1 ' ⎨ R min = 1 (σ − σ )
m

1 2 3
− Equation (I) : C1 = (σ2 + σ3 ), 0 =⇒ 2
d

2 ⎩ R max = |σ − a |
uu

 
e

1 1 1
X Th

a1
er
tin
on

.O
C

Figure 4.24: Initial limitation of the feasible domain.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
156 C HAPTER 4. S TRESS

rs
ee
s gin
Figure 4.25: Feasible region.

t d le En

r

ba
ge ro or
&1 ' ⎨ R min = 1 (σ − σ )

eS m
ci
2 1 3
− Equation (II) : C2 = (σ1 + σ3 ), 0 =⇒
f 2

ra
2 ⎩ R max = |σ − a |
C d P cs
 
b
a
2 2 2
a2
i
an an n


y ha

&1 ' ⎨ R min = 1 (σ − σ )


le
3 1 2
− Equation (III) : C3 = (σ1 + σ2 ), 0 =⇒ 2
liv or ec

2 ⎩ R max = |σ − a |


M

.A

3 3 3
a3
For each case, a feasible region that consists in a semi-annulus defined by the
m

minimum and maximum radii is obtained. Obviously, the final feasible region
uu
e

must be in the intersection of these semi-annuli, as depicted in Figure 4.25.


X Th

er

Figure 4.26 shows the final construction that results of the three Mohr’s semi-
tin

circles that contain points σ1 , σ2 and σ3 . It can also be shown that every point
on

.O

within the domain enclosed by the Mohr’s circles is feasible (in the sense that
the corresponding values of σ and τ correspond to stress states on a certain plane
C

that contains point P).


©

The construction of Mohr’s circle is trivial (once the three principal stresses
are known) and is useful for discriminating possible stress states on planes, de-
termining maximum values of shear stresses, etc.

Figure 4.26: Mohr’s circle in three dimensions.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Mohr’s Circle in 2 Dimensions 157

Example 4.3 – The principal stresses at a certain point in a continuous


medium are
σ1 = 10 , σ2 = 5 and σ3 = 2 .
The normal and tangential stresses on a plane that contains this point are σ
and τ, respectively. Justify if the following values of σ and τ are possible or
not.
a) σ = 10 and τ = 1.
b) σ = 5 and τ = 4.
c) σ = 3 and τ = 1.

rs
Solution

ee
s gin
The Mohr’s circle for the defined stress state is drawn and the given points
are marked in the half-space σ − τ.

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

Only the points belonging to the gray zone represent stress states (feasible
uu
e

points). It is verified that none of the given points are feasible.


X Th

er
tin
on

.O

4.7 Mohr’s Circle in 2 Dimensions


C

Many real-life problems in engineering are assimilated to an ideal bi-dimensional


©

stress state9 in which one of the principal stress directions is known (or assumed)
a priori. In these cases, the Cartesian axis x3 (or z-axis) is made to coincide with
said principal direction (see Figure 4.25) and, thus, the components of the stress
tensor can be written as
⎡ ⎤ ⎡ ⎤
σ11 σ12 0 σx τxy 0
not ⎢ ⎥ ⎢ ⎥
σ ≡ ⎣ σ12 σ22 0 ⎦ = ⎣ τxy σy 0 ⎦ . (4.61)
0 0 σ33 0 0 σz

9 This type of problems will be analyzed in depth in Chapter 7, dedicated to bi-dimensional


elasticity.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
158 C HAPTER 4. S TRESS

Consider now only the family of planes parallel to the x3 -axis (therefore, the
component n3 of its unit normal vector is null). The corresponding traction vec-
tor is ⎡ ⎤ ⎡ ⎤⎡ ⎤
t1 σ11 σ12 0 n1
⎢ ⎥ ⎢ ⎥⎢ ⎥
t (P, n) = σ · n =⇒ ⎣ t2 ⎦ = ⎣ σ12 σ22 0 ⎦ ⎣ n2 ⎦ (4.62)
0 0 0 σ33 0

and its component t3 vanishes. In (4.61) and (4.62) the components of the stress
tensor, σ , of the unit normal vector defining the plane, n, and of the traction

rs
vector, t, associated with direction x3 are either well known (this is the case for
σ13 , σ23 , n3 or t3 ), or do not intervene in the problem (as is the case for σ33 ). This

ee
circumstance suggests ignoring the third dimension and reducing the analysis to

s gin
the two dimensions associated with the x1 - and x2 -axes (or x- and y-axes), as
indicated in Figure 4.27. Then, the problem can be defined in the plane through

t d le En
the components of the stress tensor
( ) ( )

r
σ11 σ12 σx τxy

ba
ge ro or
eS m
not
σ≡ = (4.63)

ci
σ12 σ22 τxy σy
f

ra
C d P cs
b
a
and the components of the traction vector
i
an an n

( ) ( )( )
y ha

not t1 σ11 σ12 n1


t (P, n) = σ · n ≡ = . (4.64)
le
liv or ec

t2 σ12 σ22 n2
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 4.27: Reduction of the problem from three to two dimensions.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Mohr’s Circle in 2 Dimensions 159

4.7.1 Stress State on a Given Plane


Consider a plane (always parallel to the z-axis) whose unit normal vector n forms
an angle θ with the x-axis. A unit vector m is defined in the tangential direction
to the trace of the plane as indicated in Figure 4.28.

Remark 4.17. The unit normal vector n, the unit tangent vector m,
and the angle θ in Figure 4.28 have the following positive directions
associated with them.

rs
• Unit normal vector n: towards the exterior of the plane (with re-

ee
spect to the position of point P).

s gin
• Unit tangent vector m: generates a clockwise rotation with re-
spect to point P.

t d le En
• Angle θ : defined as counterclockwise.

r
ba
ge ro or
eS m
ci
f

ra
Consider σ , the stress tensor at a given point, whose components are defined
C d P cs
b
a
in a Cartesian base, ( )
i
an an n

σx τxy
y ha

not
σ≡ . (4.65)
τxy σy
le
liv or ec

Using (4.64), the traction vector on the given point, which belongs to the plane
M

.A

considered, is
m

( )( ) ( )
d

σx τxy cos θ σx cos θ + τxy sin θ


uu

not
t = σ ·n ≡ = .
e

(4.66)
X Th

τxy σy sin θ τxy cos θ + σy sin θ


er
tin
on

.O
C

Figure 4.28: Stress state on a given plane.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
160 C HAPTER 4. S TRESS

Taking into consideration the expression t = σθ n + τθ m, the normal stress


σθ and the tangent stress τθ on the plane with inclination θ (see Figure 4.28)
are defined, respectively, as
 
not cos θ
σθ = t · n ≡ [σx cos θ + τxy sin θ , τxy cos θ + σy sin θ ] =
sin θ (4.67)
= σx cos2 θ + τxy 2 sin θ cos θ + σy sin2 θ

and
 

rs
not sin θ
τθ = t · m ≡ [σx cos θ + τxy sin θ , τxy cos θ + σy sin θ ] =

ee
− cos θ (4.68)

2

s gin
= σx sin θ cos θ − σy sin θ cos θ + τxy sin θ − cos θ ,2

t d le En
which can be rewritten as10

r
ba
ge ro or
eS m
σx + σy σx − σy

ci
σθ = + cos (2θ ) + τxy sin (2θ )
f

ra
2 2
C d P cs
(4.69)
b
a
σx − σy
i
an an n

τθ = sin (2θ ) − τxy cos (2θ )


y ha

2
le
liv or ec
M

.A

Direct problem
m

d
uu
e
X Th

er
tin
on

.O
C

Inverse problem

Figure 4.29: Direct and inverse problems.

10 The following trigonometric relations are used here: sin (2θ ) = 2 sin θ cos θ ,
cos2 θ = (1 + cos (2θ ) ) / 2 and sin2 θ = (1 − cos (2θ ) ) / 2.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Mohr’s Circle in 2 Dimensions 161

4.7.2 Direct Problem: Diagonalization of the Stress Tensor


The direct problem consists in obtaining the principal stresses and the principal
stress directions given the components of the stress tensor (4.65) in a certain
system of axes x − y (see Figure 4.29).
The principal stress directions associated with the x - and y -axes defined by
the angles α and π/2 + α (see Figure 4.29) determine the inclinations of the
two planes on which the stresses only have a normal component σα , being the
tangent component τα null. Imposing this condition on (4.69) yields

σx − σy τxy

rs
τα = sin (2α) − τxy cos (2α) = 0 =⇒ tan (2α) = σ − σ ,

ee
2 x y
2

s gin
1 τxy
sin (2α) = ± % = ± * ,

t d le En
1 2
1+ 2 σx − σy
tan (2α) + τxy
2

ba
(4.70)

ge ro or
2

eS m
ci
f

ra
σx − σy
C d P cs
b
a
1
i
cos (2α) = ± + = ± * 2 .
an an n

1 + tan2 (2α) 2
y ha

σx − σy
+ τxy
2
le

liv or ec

2
M

.A

Equation (4.70) provides two solutions (associated with the + and − signs) α1
m

and α2 = α1 + π/2, which define the two principal stress directions (orthogonal)
d
uu

to the plane being analyzed11 . The corresponding principal stress directions are
e
X Th

obtained replacing the angle θ = α in (4.70) in (4.69), resulting in


er
tin

σx + σy σx − σy
on

.O

σα = + cos (2α) + τxy sin (2α) . (4.71)


2 2
C


©

* 



⎪ σ x + σy σx − σy 2

⎨ σ1 = + + τxy
2
2 2
σα → * (4.72)

⎪ 2

⎪ σ + σ σ − σ

⎩ σ2 =
x y

x y
+ τxy
2
2 2

11 The third principal stress direction is the direction perpendicular to the plane being ana-
lyzed (z- or x3 -axis), see (4.61) and Figure 4.27.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
162 C HAPTER 4. S TRESS

rs
Figure 4.30: Inverse problem.

ee
s gin
4.7.3 Inverse Problem

t d le En
The problem consists in obtaining the stress state on any plane given the prin-
cipal stresses and the principal stress directions σ1 and σ2 in the plane being

r
ba
analyzed. The stress state on any plane is characterized by the angle β that

ge ro or
eS m
ci
forms the unit normal vector of the plane with the principal stress direction cor-
f

ra
responding to σ1 . As a particular case, the components of the stress tensor on
C d P cs
b
a
an elemental rectangle associated with the system of axes x − y can be obtained
i
an an n

(see Figure 4.29).


y ha

Consider now the Cartesian system x − y , associated with the principal stress
le
directions (see Figure 4.30). Applying (4.69) with σx = σ1 , σy = σ2 , τx y = 0
liv or ec

and θ ≡ β results in
M

.A
m

σ1 + σ2 σ1 − σ2
d

σβ = + cos (2β )
uu
e

2 2 (4.73)
X Th

er
tin

σ1 − σ2
τβ = sin (2β )
2
on

.O
C

4.7.4 Mohr’s Circle for Plane States (in 2 Dimensions)


Consider all the possible planes that contain point P and the values of the normal
and tangent stresses, σθ and τθ , defined in (4.69) for all the possible values of
θ ∈ [0, 2π]. The stress state in the point for an inclined plane θ can now be
characterized by means of the pair

(σ = σθ , τ = τθ ) where σ ∈ R and τ ∈R, (4.74)


which, in turn, determines a point (x ≡ σ , y ≡ τ) ∈ R × R of the plane σ − τ
in Figure 4.31. To determine the locus of points of said plane that characterizes

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Mohr’s Circle in 2 Dimensions 163

all the possible stress states for planes that contain the point being analyzed, the
ensuing procedure is followed.
Considering a reference system that coincides with the principal stress di-
rections (as in Figure 4.30) and characterizing the inclination of the planes
by means of the angle β with the principal stress direction σ1 , one obtains
from (4.73) ⎧

⎪ σ + σ2 σ1 − σ2
⎨σ − 1 = cos (2β )
2 2 (4.75)

⎩ τ = σ1 − σ2 sin (2β )

rs
2

ee
and, squaring both equations and adding them up results in
 

s gin
σ1 + σ2 2 σ1 − σ 2 2
σ− +τ =2
. (4.76)

t d le En
2 2

r
Note that this equation, which will be valid for any value of the angle β , or,

ba
ge ro or
eS m
in other words, for any arbitrarily oriented plane that contains the point, corre-

ci
f
sponds to a circle with center C and radius R in the plane σ − τ given by (see

ra
C d P cs
b
a
Figure 4.31) 
i
σ1 + σ2 σ1 − σ2
an an n

C= ,0 and R = . (4.77)
y ha

2 2
le
liv or ec

Consequently, the locus of points representative of a stress state on the planes


that contain P is a circle (named Mohr’s circle), whose construction is defined
M

.A

in Figure 4.31.
m

The inverse proposition is also true: given a point of Mohr’s circle with co-
d

ordinates (σ , τ), there exists a plane that contains P whose normal and tangent
uu
e

stresses are σ and τ, respectively. In effect, using (4.75) the following trigono-
X Th

er
tin

metric expressions are obtained.


on

.O
C

Figure 4.31: Mohr’s circle for plane stress states.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
164 C HAPTER 4. S TRESS

rs
ee
Figure 4.32: Interpretation of the angle β .

s gin


t d le En
σ1 + σ2
σ−
2 σ −a

r
cos (2β ) =  =

ba
ge ro or
eS m
σ1 − σ2 R

ci
f

ra
2 (4.78)
C d P cs
b
a
τ τ
i
sin (2β ) = =
an an n

σ1 − σ R
y ha

2
le
2
liv or ec
M

These expressions uniquely define the angle β between the normal direction to
.A

the plane and the principal stress direction σ1 . The plane obtained corresponds
m

to the aforementioned stresses σ and τ. Figure 4.32 provides an interpretation


d
uu

of the angle 2β in the Mohr’s circle itself.


e
X Th

er
tin

4.7.5 Properties of the Mohr’s Circle


on

.O

a) Obtaining the point in Mohr’s circle that is representative of the stress state
C

on a plane whose normal direction forms an angle β with the principal stress
©

direction σ1 .
Take a representative point of the plane on which the principal stress direc-
tion σ1 acts (point (σ1 , 0)) and rotate an angle 2β in the direction going
from σ1 to σβ (see Figure 4.32 and Figure 4.33).

b) The representative points in Mohr’s circle of two orthogonal planes are


aligned with the center of the circle (as a consequence of property a) ) for
β2 = β1 + π/2 (see Figure 4.34).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Mohr’s Circle in 2 Dimensions 165

rs
ee
s gin
t d le En
Figure 4.33: Representative point associated with angle β in Mohr’s circle.

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

Figure 4.34: Representative points for two orthogonal planes in Mohr’s circle.
d
uu
e
X Th

er
tin

c) Mohr’s circle can be drawn if the stress state on two orthogonal planes is
known.
on

.O

In effect, by means of property b) the points representative of these two or-


C

thogonal planes in plane σ − τ are aligned with the center of Mohr’s circle.
©

Therefore, joining both points provides the intersection with the σ -axis that
corresponds to the center of the circle. Since two additional points of the
circle are known, the circle can be drawn.

d) Mohr’s circle can be drawn if the components of the stress tensor in a certain
orthonormal base are known.
This is a particular case of property c) in which the points representative
of a stress state on Cartesian planes are known (see Figure 4.35). Note, in
this figure, how the radius and the diametrical points of the circle can be
obtained. In addition, note that the application of property a) on the point
representative of the plane perpendicular to the x-axis implies moving in the

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
166 C HAPTER 4. S TRESS

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
Figure 4.35: Calculation of the radius and diametrical points of Mohr’s circle for a stress
an an n
y ha

state on Cartesian planes.


le
liv or ec
M

.A

opposite direction to that of angle α (angle of σx with σ1 = - angle of σ1 with


σx = −α).
m

d
uu
e

4.7.6 The Pole of Mohr’s Circle


X Th

er
tin
on

.O

Theorem 4.1. There exists a point in Mohr’s circle denoted pole or


C

origin of planes that has the following properties:


©

• Any straight line drawn from the pole P will intersect Mohr’s cir-
cle at a point A that represents the stress state on a plane parallel
in space to that line (see Figure 4.36).
• The inverse is also verified, that is, if a straight line, paral-
lel to a given plane, is drawn from the pole P, the intersection
point B represents the stress state on this particular plane (see
Figure 4.37).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Mohr’s Circle in 2 Dimensions 167

rs
ee
s gin
Figure 4.36: First property of the pole of Mohr’s circle.

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

Figure 4.37: Second property of the pole of Mohr’s circle.


er
tin
on

.O

Proof
C

Consider the stress tensor at the point being analyzed and its graphical rep-
resentation on the Cartesian planes of Figure 4.38 (left)12 denoted as plane A
(vertical plane) and plane B (horizontal plane). A and B are the corresponding
points in the Mohr’s circle drawn in Figure 4.38 (right).
1) Assuming property a) is verified, the pole of Mohr’s circle can be obtained
by drawing a vertical line from point A (parallel to plane A). Then, the pole P
is located at the intersection of this line with the Mohr’s circle. Also, drawing
a horizontal line from point B (parallel to plane B) determines the location of

12Note that, following the sign criterion of Mohr’s circle, the tangent stress on plane A is
τ = −τxy .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
168 C HAPTER 4. S TRESS

rs
ee
Figure 4.38: Proof of the properties of the pole of Mohr’s circle (1).

s gin
t d le En
the pole at the intersection of this line with the Mohr’s circle. The same point
P is obtained in both cases, as is verified in the Figure 4.38.

r
ba
ge ro or
eS m
2) Consider now an arbitrary plane whose normal direction forms an angle θ

ci
f

ra
with the horizontal direction (see Figure 4.39, left) and consider also the
C d P cs
normal and tangent stresses, σθ and τθ , respectively, according to this plane.
b
a
i
Assuming that the major principal stress direction σ1 forms an angle α with
an an n

the direction of stress σx , then, the direction of stress σθ forms an angle


y ha

(θ − α) with the major principal stress direction σ1 .


le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 4.39: Proof of the properties of the pole of Mohr’s circle (2).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Mohr’s Circle in 2 Dimensions 169

3) Consider the Mohr’s circle and the pole P obtained in step 1) (see Figure 4.39,
right)13 . Using property a) of Section 4.7.5, point C can be obtained. This
point is representative of the Mohr’s circle that corresponds to the plane con-
sidered, obtained by rotating from point M a double angle equal to 2 (θ − α)
such that the angle MOC is 2 (θ − α). By construction, angle AOM is 2α
and angle AOC, the sum of both, is 2 (θ − α) + 2α = 2θ . The arc included
by this angle is AMC = 2θ . Then, the angle semi-inscribed in APC, which
includes arc AMC, will be θ , which proves that the straight line PC is paral-
lel to the trace of the plane considered. Since this plane could be any plane,
the validity of the property is proven.

rs
ee
Example 4.4 – Calculate the stresses acting on state III = I + II:

s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

Solution
d
uu
e

To be able to add states I and II, the stresses must act on the same planes.
X Th

er
tin

Since the two states present planes with different orientations, the stresses
acting in state II must be found for the planes given in state I. To this aim, the
on

.O

Mohr’s circle for state II must be drawn.


C

13 The following geometric properties are used here: a) the value of a central angle of a circle
is the same as the arc it includes; and b) the value of an angle semi-inscribed in a circle is
equal to half the arc it includes.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
170 C HAPTER 4. S TRESS

rs
ee
s gin
t d le En
To draw the circle, planes a and b are represented since their stress states are
known. The corresponding points in the Mohr’s circle belong to the abscissa

r
ba
and determine, thus, the diameter of the circle.

ge ro or
eS m
The pole is obtained as the intersection of the lines that are parallel to the two

ci
f
planes inclined at 45◦ and that contain the points that they represent. Once

ra
C d P cs
b
a
the pole is determined, a horizontal line is drawn from it, whose intersection
i
an an n

with the Mohr’s circle (because it is tangent to the point, the intersection in
y ha

this case is the same pole) determines the point representative of the horizon-
tal plane (2, 1). The same procedure is repeated for a vertical plane to obtain
le
liv or ec

point (2, −1). With this information, state II can be reconstructed on the hor-
M

.A

izontal and vertical planes. Then, the stresses obtained are added to those of
state I to finally obtain state III.
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Mohr’s Circle in 2 Dimensions 171

continuum mechanics soil mechanics


Figure 4.40: Differences in the sign criterion for continuum mechanics and soil mechan-
ics.

rs
ee
s gin
4.7.7 Mohr’s Circle with the Soil Mechanics Sign Criterion
The sign criterion, with respect to the normal and tangent stresses, used in soil

t d le En
mechanics is the inverse of the one used in continuum mechanics (see Fig-

r
ure 4.40). The differences are:

ba
ge ro or
eS m
ci
• The positive stresses in soil mechanics are in the opposite direction (normal
f

ra
C d P cs
stresses are positive when they are compressive, and the direction of the pos-
b
a
itive tangent stresses is defined by a counterclockwise rotation with respect
i
an an n

to the plane).
y ha

• The sign criterion for angles is the same (counterclockwise angles are posi-
le
liv or ec

tive).
M

.A

Consequently, if the order of the principal stresses is respected (σ1 ≥ σ2 ), the


order of the principal stresses will be inverted in soil mechanics with respect to
m

continuum mechanics for a same stress state (see Figure 4.41).


uu
e

Consider the fundamental expressions in (4.73), which are the starting point
X Th

er
tin

in the construction and determination of the properties of the Mohr’s circle.


Using the two sign criteria for a same stress state results in:
on

.O
C

continuum mechanics soil mechanics


Figure 4.41: Direction of the principal stresses for continuum mechanics and soil me-
chanics.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
172 C HAPTER 4. S TRESS

Continuum mechanics: σβ , τβ , σ1 , σ2 , β
⎧ ∗

⎪ σβ = −σβ


⎪ ∗
⎨ τβ = −τβ (4.79)
Soil mechanics: σ1∗ = −σ2



⎪ σ ∗ = −σ1

⎩ 2∗
β = β + π/2

Replacing (4.79) in (4.73) yields


−σ2∗ − σ1∗ −σ2∗ + σ1∗

rs
−σβ∗ = + cos (2β ∗ − π) ,
 

ee
2 2
− cos (2β ∗ )

s gin
−σ2∗ + σ1∗ (4.80)
−τβ∗ = sin (2β ∗ − π) ,
 

t d le En
2
− sin (2β ∗ )

r
ba
ge ro or
eS m
and, operating on these expressions finally results in

ci
f

ra
σ1∗ + σ2∗ σ1∗ − σ2∗
C d P cs
σβ∗ =
b cos (2β ∗ ) ,

a
+
i
2 2
an an n

(4.81)
σ ∗ − σ2∗
y ha

τβ∗ = 1 sin (2β ∗ ) .


le
2
liv or ec

Note that the fundamental expressions in (4.81), obtained on the basis of the sign
M

.A

criterion in soil mechanics, are the same as those in (4.73), obtained on the basis
m

of the sign criterion in continuum mechanics. Therefore, the construction of the


d

Mohr’s circle and the determination of its properties is the same in both cases.
uu
e
X Th

er
tin

4.8 Mohr’s Circle for Particular Cases


on

.O

4.8.1 Hydrostatic Stress State


C

In an hydrostatic stress state, characterized by σ1 = σ2 = σ3 = σ , the Mohr’s


©

circles in three dimensions collapses into a point (see Figure 4.42).

Figure 4.42: Mohr’s circle for a hydrostatic stress state.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Mohr’s Circle for Particular Cases 173

4.8.2 Mohr’s Circles for a Tensor and its Deviator


The Mohr’s circles in three dimensions associated with a stress state and its
deviator differ in a translation equal to the mean stress (see Figure 4.43).

⎡ ⎤ ⎧
σm 0 0 ⎪ 
⎨ σ1 = σm + σ1
not ⎢ ⎥
σ= σ sph + σ  ; σ sph ≡ ⎣ 0 σm 0 ⎦ =⇒ σ2 = σm + σ2 
  ⎪

spherical deviator 0 0 σm σ3 = σm + σ3 
part part

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
Figure 4.43: Mohr’s circle for a stress state and its deviator.

ci
f

ra
C d P cs
b
a
i
an an n

4.8.3 Mohr’s Circles for a Plane Pure Shear Stress State


y ha

le
liv or ec
M

.A

Definition 4.9. A plane pure shear stress state occurs at a point


when there are two orthogonal planes on which there is only tan-
m

gent (shear) stress (see Figure 4.44).


uu
e
X Th

er
tin

The Mohr’s circle corresponding to a pure shear stress state characterized by


on

.O

a tangent stress τ ∗ has as center the origin of axes and as radius R = |τ ∗ |.


C

The proof is immediate from the construction criteria of the Mohr’s circle (see
©

Figure 4.44, left).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
174 C HAPTER 4. S TRESS

rs
Figure 4.44: Mohr’s circle for a plane pure shear stress state.

ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 175

P ROBLEMS

Problem 4.1 – The solid below is subjected to the following stress state in
equilibrium.

( )
xy 5y

rs
not
σ≡ (in MPa)

ee
5y 4x

s gin
t d le En

r
ba
ge ro or
eS m
ci
Determine:
f

ra
C d P cs
1) The expression of the forces per unit of mass acting on the solid.
b
a
i
2) The expression of the normal and tangent components of the forces act-
an an n

ing on the boundary, indicating their sign according to the Mohr’s cir-
y ha

cle criterion.
le
liv or ec
M

.A
m

Solution
d
uu
e

1) The expression of the body forces is obtained directly from the internal equi-
X Th

er

librium equation (4.26),


tin

 ( ) ( )
on

.O

1 not 1 ∂ ∂ xy 5y 1 y+5
b = − ∇ · σ =⇒ b ≡ − , =− .
ρ ρ ∂x ∂y ρ
C

5y 4x 0
©

2) The normal (σ ) and tangent (τ) components of the body forces acting on the
boundary are given by

σ = t·n and τ = t·m with t = n · σ ,


where n and m are the unit normal vector and the unit tangent vector of the
boundary, respectively. The boundary of the solid can be divided into three parts,
according to their n and m vectors:

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
176 C HAPTER 4. S TRESS

rs
ee
Boundary 1
The traction vector for this surface is

s gin
( )( )
1 xy 5y 1 xy + 5y

t d le En
not
t1 = n1 · σ ≡ √ [1, 1] =√ .
2 5y 4x 2 5y + 4x

r
ba
ge ro or
eS m
ci
Then, the corresponding normal and tangent components of the body forces are
f

ra
( )
C d P cs
b
a
not 1 1 1 1
i
σ1 = t1 · n1 ≡ √ [xy + 5y, 5y + 4x] √ = (4x + 10y + xy) ,
an an n

2 2 1 2
y ha

( )
le
liv or ec

not 1 1 1 1
τ1 = t1 · m1 ≡ √ [xy + 5y, 5y + 4x] √ = (−4x + xy) .
M

.A

2 2 −1 2
m

This is now particularized for the x and y values corresponding to the boundary,
d
uu

that is, for y = 1 − x and x ∈ [0, 1],


e
X Th

er


tin


⎪ 1

⎨ σ1 = 10 − 5x − x2 with x ∈ [0, 1] ,
on

.O

2

⎩ τ1 = 1
−3x − x2 with x ∈ [0, 1] .

C

Boundary 2
The traction vector for this surface is
( ) ( )
not xy 5y −5y
t2 = n2 · σ ≡ [0, −1] = .
5y 4x −4x

Then, the corresponding normal and tangent components of the body forces are

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 177

( )
not 0
σ2 = t2 · n2 ≡ [−5y, −4x] = 4x ,
−1
( )
not −1
τ2 = t2 · m2 ≡ [−5y, −4x] = 5y .
0
This is now particularized for the x and y values corresponding to the boundary,
that is, for y = 0 and x ∈ [0, 1],

⎨ σ2 = 4x with x ∈ [0, 1] ,

rs
ee
⎩τ = 0.
2

s gin
t d le En
Boundary 3

r
The traction vector for this surface is

ba
ge ro or
( ) ( )

eS m
ci
−xy
f xy 5y

ra
not
t3 = n3 · σ ≡ [−1, 0] = .
C d P cs
b −5y

a
5y 4x
i
an an n
y ha

Then, the corresponding normal and tangent components of the body forces are
( )
le
liv or ec

not −1
σ3 = t3 · n3 ≡ [−xy, −5y] = xy ,
M

.A

0
m

( )
d
uu

0
e

not
τ3 = t3 · m3 ≡ [−xy, −5y] = −5y .
X Th

er

1
tin

This is now particularized for the x and y values corresponding to the boundary,
on

.O

that is, for x = 0 and y ∈ [0, 1],


C


©

⎨ σ3 = 0 ,
⎩ τ = −5y with y ∈ [0, 1] .
3

Note that the results for boundaries 2 and 3 could have been obtained by direct
comparison since they are a horizontal and a vertical surface, respectively:

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
178 C HAPTER 4. S TRESS

rs
ee
s gin
⎧ ⎧
⎨ σ2 = σy with x ∈ [0, 1] ⎨ σ3 = σx with x = 0

t d le En
⎩τ = τ with y = 0 ⎩ τ = −τ = −5y with y ∈ [0, 1]

r
2 xy 3 xy

ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
Finally, the expression of the normal and tangent components of the forces act-
i
an an n

ing on the boundary of the solid are drawn, indicating the most significant val-
y ha

ues.
le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 179

Problem 4.2 – The following is known of a stress state.


1) The z-direction is a principal stress direction and σzz = a.
2) The mean stress is σm = a > 0.
3) The maximum shear stress in the planes that are parallel to the z-axis
is τmax = b > 0.
Draw, indicating the most significant values, the Mohr’s circle in three dimen-
sions of the stress tensor and its deviatoric tensor.

rs
ee
s gin
Solution

t d le En
Note that the only difference there will be between the two circles is that one
will be translated a distance σm with respect to the other.

r
ba
By means of the definition of the deviatoric stress tensor,

ge ro or
eS m
ci
σ  = σ − σm 1 =⇒ f
σzz = σzz − σm = a − a = 0 =⇒ σzz = 0

ra
C d P cs
b
a
i
is deduced. The fact that the trace is an invariant and that the trace of the devia-
an an n

σ  ) = 0, results in
toric stress tensor is zero, Tr (σ
y ha


le
liv or ec

  σzz = σ2 = 0 ,
σxx + σyy = 0 =⇒
M

.A

σ1 + σ3 = 0 .
m

Finally, the radius of the major circumference (between σ1 and σ3 ) is determined
d
uu
e

through the application of condition 3). The two Mohr’s circles are shown below.
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
180 C HAPTER 4. S TRESS

Problem 4.3 – Given the following information of a stress state in a certain


point,
1) σx = 1 (where the x-axis is a principal stress direction).
2) The maximum shear stress in the planes that are parallel to the x-axis
is 3.
3) The maximum shear stress in the planes that are parallel to the minor
principal stress direction is 2.

rs
obtain all the possible Mohr’s circles corresponding to this state, indicating the
values of the principal stresses.

ee
s gin
t d le En
Solution

r
ba
The following property of the Mohr’s circle in 3D must be taken into account to

ge ro or
eS m
solve this problem.

ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Circle number:
1 − corresponds to planes parallel to the principal stress direction of σ3 .
2 − corresponds to planes parallel to the principal stress direction of σ1 .
3 − corresponds to planes parallel to the principal stress direction of σ2 .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 181

Then, the following possibilities are considered.


1. σx is the major principal stress, which results in the following Mohr’s
circle.

rs
ee
s gin
t d le En
2. σx is the intermediate principal stress, which results in the following

r
ba
ge ro or
Mohr’s circle.

eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

3. σx is the minor principal stress, which is an impossible situation because


er
tin

conditions 2) and 3) cannot be satisfied at the same time since they refer
on

to the maximum shear stress on the same plane.


.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
182 C HAPTER 4. S TRESS

Problem 4.4 – Determine the values of α and β for which the following stress
states are possible, considering that σ > 0 and τ = 0.5σ .

rs
ee
s gin
t d le En

r
Solution

ba
ge ro or
eS m
ci
f
The problem is solved following the same steps in all three cases, which are:

ra
C d P cs
b
a
Step 1: Draw the Mohr’s circle corresponding to the stress state. Even for the
i
stress states in which only two different pairs of points (σ , τ) belonging to the
an an n
y ha

Mohr’s circle are given, the circle can be drawn taking into account that it must
le
be symmetric with respect to the longitudinal axis.
liv or ec

Step 2: Identify the pole. In all cases, a straight horizontal line is drawn, which
M

.A

must contain the point of the Mohr’s circle corresponding to the horizontal plane.
m

Then, the pole is identified as the point where the line crosses the circle again.
d

The horizontal plane is used to identify the pole because, of the three planes
uu
e

shown for each stress state, it is the only one with a known orientation.
X Th

er
tin

Step 3: Draw a straight line joining the pole and the two (σ , τ) points corre-
on

sponding to the planes whose inclination must be obtained. The inclination of


.O

these planes and, thus, the angles α and β are given directly by the orientation
C

of the lines drawn.


©

Step 4: The schematic description of the stress states on the three planes can be
redrawn with the appropriate inclination.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 183

(a)

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

(b)
er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
184 C HAPTER 4. S TRESS

rs
ee
s gin
(c)

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 185

Problem 4.5 – Calculate the possible values of σ , σ  , σ  , τ, τ  and α for which


state III is the sum of states I and II, considering that τ ≥ 0.

rs
ee
s gin
t d le En

r
Solution

ba
ge ro or
eS m
ci
Stress state II on the vertical plane must be found to be able to add states I and
f

ra
II together.
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O

The Mohr’s circle of state II will allow determining the normal and shear stress
C

on the vertical plane. The known stress state on the horizontal plane (4, −3)
©

belongs to the Mohr’s circle. Since it is known to be symmetric with respect


to the longitudinal axis, the stress state (4, 3) must also belong to the Mohr’s
circle. Observing the figure representing state II, and considering that τ ≥ 0, it
is concluded that this point, (4, 3), corresponds to the stress state on the plane
inclined at 45◦ in the counterclockwise direction. Thus,

τ =3 .

Now, a third point belonging to the Mohr’s circle must be obtained in order to be
able to draw the complete circle. Because there exists only one pole and it must
belong to the Mohr’s circle, finding this point will allow completing the circle. A
straight horizontal line (parallel to the horizontal plane) is draw at point (4, −3),

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
186 C HAPTER 4. S TRESS

which corresponds to the stress state on a horizontal plane. Another straight line,
parallel to the other plane with a known stress state, the plane inclined at 45◦
in the counterclockwise direction, is drawn passing through the corresponding
stress state, (4, 3). The point where these to lines meet provide the pole of the
Mohr’s circle, which is found to be at (−2, −3):

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

Once these three points are known, the Mohr’s circle can be drawn. Before cal-
le
liv or ec

culating the stress state on the vertical plane, the value of σ is sought. To obtain
the stress state on the plane inclined at 45◦ in the clockwise direction, a straight
M

.A

line must be drawn, parallel to this plane, that crosses the pole.
m

d
uu
e
X Th

er
tin
on

.O
C

This results in a line tangent to the pole, therefore, the stress state corresponding
to the pole is also the stress state on this plane and

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 187

σ = −2 .
Finally, a vertical line is drawn from the pole and the intersection of this line with
the Mohr’s circle provides the stress state on the vertical plane, which results
in (−2, 3).

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n

Then, stress state II is defined on a vertical and horizontal plane as follows.


y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

This allows adding stress states I and II to obtain state III,


©

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
188 C HAPTER 4. S TRESS

revealing the values of σ  and σ  .

σ = 3
σ  = 7

The values of τ  and α remain to be found. To this aim, the Mohr’s circle of
stress state III must be drawn. The points corresponding to the known stress
states on the vertical and horizontal planes are marked on the σ − τ space and,
in a procedure analogous to the one used for the Mohr’s circle of state II, the
pole is obtained. The circle can now be drawn and simple trigonometry allows

rs
calculating its center at (2, 0), which will be useful in the calculation of τ  and

ee
α.

s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin

Drawing a vertical line at σ = 6 provides the values of τ  at the intersection of


this line with the circle. Two options are possible, one corresponding to a posi-
on

.O

tive value of τ  and another corresponding to the same value but with a negative
C

sign. Following the sign criterion for the Mohr’s circle, and to be consistent with
the directions drawn in the figure representing state III, the value of τ  must be
©


τ  = −3 2 .

Since there are two possible values of τ  , two values of α will exist, each cor-
responding to one of the τ  values. To obtain the values of α, a straight line is
drawn from the pole to each of the points representing the possible stress states
of the plane inclined at α in a clockwise direction.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 189

rs
ee
s gin
t d le En

r
Determining the inclination of these two lines will result directly in the possible

ba
ge ro or
eS m
values of α.

ci
f

ra
C d P cs
√ & √ '
b
a
τ  = −3 2 ⇒ α + = 180◦ − arctan 1+3 2  141◦
i
an an n

&√ '
y ha


−τ  = 3 2 ⇒ α − = arctan 2−1  8◦
le
liv or ec

3
M

.A

The two possible configurations of stress state III are pictured below.
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
190 C HAPTER 4. S TRESS

E XERCISES

4.1 – Determine all the possible values of


σ (σ > 0) and τ (τ > 0) in the figure knowing
that the maximum shear stress on any plane at
the point is τmax = 1.

rs
ee
s gin
4.2 – The following is known of the stress state in a point of a continuous

t d le En
medium. The maximum shear stress in planes parallel to the principal stress

r
direction of σ1 is τmax = 2. Obtain all the values of σ1 , σ2 and σ3 that make

ba
ge ro or
eS m
possible the stress state σ = 2 and τ = 2 on a certain plane for the following

ci
cases (separately). f

ra
C d P cs
b
a
a) The maximum shear stress in planes parallel to the principal stress di-
i
an an n

rection of σ2 is τ2max = 2.
y ha

b) The maximum shear stress in planes parallel to the principal stress di-
le
liv or ec

rection of σ3 is τ3max = 0.
M

.A

c) The maximum shear stress in planes parallel to the principal stress di-
rection of σ2 is τ2max = 4.
m

d
uu
e
X Th

er
tin

4.3 – Determine for which values of σ ∗ the fol-


lowing stress states are possible in the planes
on

.O

belonging to P.
C

a) σ = 4 and τ = 2.
©

b) σ = 4 and τ = 1.
c) σ = 7 and τ = 0.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 191

4.4 – Obtain, in terms of τ, the principal stresses and the value of the maximum
shear stress of the state that results from the sum of states I and II.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
4.5 – Given states I and II, determine the possible values of σ and τ for which

ci
f
state III = I + II verifies that the principal stress σ2 is positive and its direction

ra
C d P cs
forms a 30◦ angle with the y-axis.
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
192 C HAPTER 4. S TRESS

4.6 – Determine all the possible values of τ ∗ for which the stress state that is the
sum of states I and II verifies the following conditions (separately).

rs
ee
s gin
a) There do not exist tensile stresses on any plane.
b) There do not exist compressive stresses on any plane.

t d le En
c) The maximum shear stress (τmax ) is less than 2.

r
d) It is a pure shear stress state.

ba
ge ro or
eS m
e) It is a hydrostatic stress state.

ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
CH.5. BALANCE PRINCIPLES
Multimedia Course on Continuum Mechanics
Overview
 Balance Principles Lecture 1
 Convective Flux or Flux by Mass Transport Lecture 2 Lecture 3
 Local and Material Derivative of a Volume Integral Lecture 4
 Conservation of Mass
 Spatial Form Lecture 5
 Material Form
 Reynolds Transport Theorem
Lecture 6
 Reynolds Lemma
 General Balance Equation Lecture 7
 Linear Momentum Balance
 Global Form Lecture 8
 Local Form

2
Overview (cont’d)
 Angular Momentum Balance
 Global Spatial Lecture 9
 Local Form
 Mechanical Energy Balance
 External Mechanical Power
Lecture 10
 Mechanical Energy Balance
 External Thermal Power
 Energy Balance
 Thermodynamic Concepts Lecture 11
 First Law of Thermodynamics
 Internal Energy Balance in Local and Global Forms Lecture 12
 Second Law of Thermodynamics Lecture 13
 Reversible and Irreversible Processes
Lecture 14
 Clausius-Planck Inequality
Lecture 15

3
Overview (cont’d)
 Governing Equations
 Governing Equations
Lecture 16
 Constitutive Equations
 The Uncoupled Thermo-mechanical Problem

4
5.1. Balance Principles
Ch.5. Balance Principles

5
Balance Principles
The following principles govern the way stress and deformation vary in
the neighborhood of a point with time.
REMARK
 The conservation/balance principles: These principles are always
 Conservation of mass valid, regardless of the type of
material and the range of
 Linear momentum balance principle
displacements or deformations.
 Angular momentum balance principle
 Energy balance principle or first thermodynamic balance principle

 The restriction principle:


 Second thermodynamic law

 The mathematical expressions of these principles will be given in,


 Global (or integral) form
 Local (or strong) form

6
5.2. Convective Flux
Ch.5. Balance Principles

7
Convection
 The term convection is associated to mass transport, i.e., particle
movement.
 Properties associated to mass will be transported with the mass when
there is mass transport (particles motion) convective transport

 Convective flux of an arbitrary property A through a control


surface S :

amount of A crossing S
ΦS =
unit of time

8
Convective Flux or
Flux by Mass Transport
 Consider:
 An arbitrary property A of a continuum medium (of any tensor order)
 The description of the amount of the property per unit of mass, Ψ ( x,t )
(specific content of the property A ) .
 The volume of particles dV crossing a
differential surface dS during the
interval [t , t + dt ] is
dV =dS ⋅ dh =v ⋅ n dt dS
= ρ dV
dm = ρ v ⋅ n dSdt
 Then,
 The amount of the property crossing the differential surface per unit of
time is: Ψ dm
d ΦS = = ρ Ψ v ⋅ n dS
dt
9
Convective Flux or
Flux by Mass Transport
 Consider:
 An arbitrary property A of a
continuum medium (of any tensor order) inflow
v ⋅n ≤ 0
outflow
 The specific content of A (the amount v ⋅n ≥ 0
of A per unit of mass) Ψ ( x,t ) .

 Then,
 The convective flux of A through a spatial surface, S , with unit
normal n is:
Φ S ( t )= v is velocity

s
ρ Ψ v ⋅ n dS Where:
ρ is density

 If the surface is a closed surface, S = ∂V , the net convective flux is:


Φ ∂V ( t=
) ∫ ρ Ψ v ⋅ n dS = outflow - inflow
∂V

10
Convective Flux
REMARK 1
The convective flux through a material surface is always null.
REMARK 2
Non-convective flux (conduction, radiation). Some properties can be
transported without being associated to a certain mass of particles. Examples of
non-convective transport are: heat transfer by conduction, electric current flow,
etc.
Non-convective transport of a certain property is characterized by the non-
convective flux vector (or tensor) q ( x,t ) :
∫ q ⋅ n dS ; convective flux =
non - convective flux =
s ∫ ρψ v ⋅ n dS
s

non-convective flux convective


vector flux vector

11
Example
Compute the magnitude and the convective flux Φ S which correspond to the
following properties:
a) volume
b) mass
c) linear momentum
d) kinetic energy

12
Φ S ( t )=
Example - Solution ∫ ρ Ψ v ⋅ n dS
s

a) If the arbitrary property is the volume of the particles:


A ≡V

The magnitude “property content per unit of mass” is volume per unit of
mass, i.e., the inverse of density:

V 1
Ψ
= =
M ρ

The convective flux of the volume of the particles V through the surface S is:

1
Φ S= ∫ρ
s ρ
v ⋅ n dS= ∫
s
v ⋅ n dS VOLUME FLUX

13
Φ S ( t )=
Example - Solution ∫ ρ Ψ v ⋅ n dS
s

b) If the arbitrary property is the mass of the particles:


A ≡M

The magnitude “property per unit of mass” is mass per unit of mass, i.e., the
unit value:

M
Ψ
= = 1
M

The convective flux of the mass of the particles M through the surface S is:

Φ
= S ∫ρ
s
1 v ⋅n =
dS ∫s
ρ v ⋅ n dS MASS FLUX

14
Φ S ( t )=
Example - Solution ∫ ρ Ψ v ⋅ n dS
s

c) If the arbitrary property is the linear momentum of the particles:


A ≡M v

The magnitude “property per unit of mass” is mass times velocity per unit of
mass, i.e., velocity:

Mv
=
Ψ = v
M

The convective flux of the linear momentum of the particles M v through the
surface S is:

=ΦS ∫ ρ v ( v ⋅ n ) dS
s
MOMENTUM FLUX

15
Φ S ( t )=
Example - Solution ∫ ρ Ψ v ⋅ n dS
s

d) If the arbitrary property is the kinetic energy of the particles:


1
A ≡ M v2
2
The magnitude “property per unit of mass” is kinetic energy per unit of
mass, i.e.:
1
M v2
1 2
=Ψ 2= v
M 2
1
The convective flux of the kinetic energy of the particles M v 2 through the
2
surface S is:
1
= Φ S ∫ ρ v 2 ( v ⋅ n ) dS KINETIC ENERGY FLUX
s 2

16
5.3. Local and Material Derivative
of a Volume Integral
Ch.5. Balance Principles

17
Derivative of a Volume Integral
 Consider:
 An arbitrary property A of a continuum medium (of any tensor order)
 The description of the amount of the property per unit of volume
(density of the property A ), µ ( x,t )
REMARK
and Ψ are related
 The total amount of the property through .
in an arbitrary volume, V, is:
Q (t )
Q ( t ) = ∫ µ ( x, t ) dV
V
Q ( t + ∆t )
 The time derivative of this volume integral is:
Q ( t + ∆t ) − Q ( t )
Q′ ( t ) = lim
∆t → 0 ∆t

18
Local Derivative of a Volume Integral
 Consider: Q (t )
 The volume integral Q ( t ) = ∫ µ ( x, t ) dV
V Q ( t + ∆t ) Control
Volume, V
 The local derivative of Q ( t ) is:
local not ∂ ∫V µ ( x, t + ∆t ) dV − V∫ µ ( x, t ) dV REMARK
derivative
= ∫ µ ( x, t ) dV ∆lim
∂t V t →0 ∆t The volume is fixed in
space (control volume).
It can be computed as:

∂ Q ( t + ∆t ) − Q ( t ) ∫ µ ( x, t + ∆t ) dV − ∫ µ ( x, t ) dV
= ∫ µ ( x , t ) dV =
lim lim V = V

∂t V ∆t → 0 ∆t ∆t → 0 ∆t

∫ [µ ( x, t + ∆t ) − µ ( x, t )]dV µ ( x, t + ∆t ) − µ ( x, t ) ∂µ ( x, t )
lim=
∆t → 0
V

∆t ∫V ∆
= lim
t →0
∆ t
dV ∫V ∂t dV
∂µ  x,t 
∂t
19
Material Derivative of a Volume
Integral
 Consider:
 The volume integral Q ( t ) = ∫ µ ( x, t ) dV
V

 The material derivative of Q (t ) is: Q (t ) Q ( t + ∆t )

material
not d
µ ( x, t ) dV
dt Vt ∫≡V
derivative =

µ ( x, t + ∆t ) dV − ∫ µ ( x, t ) dV REMARK
= lim
∫ V ( t +∆t ) V (t )
The volume is mobile in space
∆t → 0 ∆t
and can move, rotate and
 It can be proven that: deform (material volume).

d ∂  ∂µ   dµ 
( ) µ dV + ∫ ∇ ⋅ ( µ v )= ∫V  ∂t + ∇ ⋅ ( µ v ) =
dt Vt∫≡V ∂t V∫ ∫V  dt + µ∇ ⋅ v  dV
µ x , t = dV dV dV
  V  
material local convective
derivative of derivative of derivative of
the integral the integral the integral

20
5.4. Conservation of Mass
Ch.5. Balance Principles

21
Principle of Mass Conservation
 It is postulated that during a motion there are neither mass
sources nor mass sinks, so the mass of a continuum body is a
conserved quantity (for any part of the body).

 The total mass M ( t ) of


the system satisfies:

( t ) M ( t + ∆t ) > 0
M=

 Where:
=M (t ) ∫ ρ ( x, t ) dV ∀∆Vt ⊂ Vt
∆Vt

=
M ( t + ∆t ) ∫∆Vt +∆t
ρ ( x, t + ∆t ) dV ∀∆Vt +∆t ⊂ Vt +∆t
22
Conservation of Mass in Spatial Form
 Conservation of mass requires that the material time derivative of
the mass M ( t ) be zero for any region of a material volume,
M ( t + ∆t ) − M ( t ) d
M ′ (=
t ) lim = ∫ ρ dV
= 0 ∀∆V ⊂ V , ∀t
∆t → 0 ∆t dt ∆V ⊂V ≡V t t

 The global or integral spatial form of mass conservation principle:


d dµ
∫ µ ( x, t )=
dV ∫( + µ∇ ⋅ v ) dV
dt Vt ≡V V
dt
d  dρ 
dt ∫∆ Vt ⊂Vt ≡V
ρ (x, t )dV
= ∫∆V ⊂V  dt + ρ∇ ⋅ v

= 0
 dV ∀∆V ⊂ V , ∀t

 By a localization process we obtain the local or differential


spatial form of mass conservation principle:
for ∆V → dV (x, t ) (localization process) CONTINUITY
EQUATION
d ρ (x, t ) ∂ρ (x, t )
+ ( ρ∇ ⋅ v )(x=
, t) + ∇ ⋅ ( ρ v )(x=
, t) 0 ∀x ∈V , ∀t
dt ∂t
23
Conservation of Mass in Material Form
d F 1 dF
 Consider the relations:  = F ∇⋅v (∇ ⋅ v ) =
 dt F dt
dV = F dV
 0

 The global or integral material form of mass conservation


principle can be rewritten as:
dρ dρ 1 dF ∂ρ ( X , t ) ∂ F ( X, t )
∫V dt + ρ∇ ⋅ v=
( ) dV ∫∆V ( dt + ρ F dt =
) dV ∫V ( ∂ t F ( X, t ) + ρ ∂ t ) dV0
0   
∂ρ ( X, t ) F dV0 ∂ [ ρ |F|]( X,t )
∂t ∂ t  


→ ∫  ρ F  ( X , t ) dV
= 0 ∀∆V0 ⊂ V0 , ∀t
∂t
0
∆V0 ⊂V0

 The local material form of mass conservation principle reads :


∂ ρ0
=  ρ F  ( X , t ) 0=
ρ t =0 ( X ) F t =0 ρ t ( X ) F t ( X ) ρ= ∀X ∈V0 , ∀t
∂t  t
Ft
=1
24
5.5. Reynolds Transport Theorem
Ch.5. Balance Principles

25

Reynolds Lemma dt
+ ρ∇ ⋅ v =0

 Consider:
 An arbitrary property A of a continuum medium (of any tensor order)
 The spatial description of the amount of the property per unit of
mass, ψ ( x,t ) (specific contents ofA )
 The amount of the property A in the continuum body at time t
for an arbitrary material volume is: Q ( t ) = ∫ ρψ dV
Vt =V

 Using the material time derivative leads to,


d d   dψ dρ 
(t )
dt Vt∫≡V ∫V  dt ∫V  dt
Q ′= ρψ dV
= ( ρψ ) + ( ρψ ) ∇ ⋅ v =
 dV ρ + ψ ( + ρ ∇ ⋅ v )  dV
dt
dψ dρ
= ρ dt + ψ dt =0
 Thus,
(continuity equation)
d dψ REYNOLDS
dt Vt∫≡V ∫V dt
ρψ dV = ρ dV
LEMMA
26
d ∂
( ) ∫ ∇ ⋅ ( µ v ) dV
dt Vt∫≡V ∂t V∫
µ = x , t dV µ dV +
V

Reynolds Transport Theorem


 The amount of the property A in the continuum body at time t for
an arbitrary fixed control volume is: Q ( t ) = ∫ ρψ dV
V
 Using the material time derivative leads to,
d ∂(ρ ψ )
∫=ρψ dV ∫ dV + ∫ ∇ ⋅ ( ρ ψ
; v ) dV
dt Vt ≡V V
∂t V
= ∫ ρ dψ dV dψ
V
dt
= ∫∂V n⋅( ρ ψ v ) dV ρ
dt
 And, introducing the Reynolds Lemma ∂V
and Divergence Theorem: dV
dψ ∂ ( ρψ )
∫ρ
V
dt
=
dV ∫V ∂t dV + ∂∫V ρ ψ v ⋅ n dS
REMARK
ê3
ê 2 ∫ ρψ
V
dV

The Divergence Theorem: ê1

∫ ∇ ⋅ v dV= ∫ n ⋅ vdS
V ∂V
= ∫ v ⋅ n dS
∂V
27
dψ ∂ ( ρψ )
∫ρ dt
=
dV ∫V ∂t dV + ∂∫V ρ ψ v ⋅ n dS
Reynolds Transport Theorem
V

 The eq. can be rewritten as:


∂ dψ REYNOLDS TRANSPORT

∂t V
ρ ψ dV= ∫ρ
V
dt
dV − ∫ ρ ψ v ⋅ n dS
∂V
THEOREM

Change due to the net outward


Change (per unit of time) of convective flux of A through the
the total amount of A . within boundary ∂V .

the control volume V at time t. ρ
dt

Rate of change of the amount of property A ∂V


integrated over all particles that are filling dV
the control volume V at time t.

ê3
ê 2 ∫ ρψ
V
dV

ê1

28
Reynolds Transport Theorem

∂ dψ REYNOLDS TRANSPORT

∂t V
ρ ψ dV= ∫ρ
V
dt
dV − ∫ ρ ψ v ⋅ n dS
∂V
THEOREM (integral form)

∂ dψ dψ

∂t V
ρ ψ dV= ∫ρ dt
dV − ∫ ρ ψ v ⋅ n dS ρ
dt
V ∂V ∂V

= ∫ ( ρ ψ ) dV = ∫ ∇ ⋅ (ρ ψ v ) dV
V
∂t V
dV
∂ dψ
∫ ∂t
ρ ψ ) dV
(= ∫ [ρ
dt
− ∇ ⋅ ( ρ ψ v )] dV ∀∆V ⊂ V ∀t
∆V ⊂V ∆V ⊂V ê 3
ê 2 ∫ ρψ
V
dV

∂ dψ ê1
(ρ =
ψ) ρ − ∇ ⋅ ( ρ ψ v ) ∀x ∈ V ∀t
∂t dt
REYNOLDS TRANSPORT
THEOREM (local form)
29
5.6. General Balance Equation
Ch.5. Balance Principles

30
General Balance Equation
 Consider:
 An arbitrary property A of a
continuum medium (of any tensor order)
 The amount of the property per
unit of mass, ψ ( x,t )
 The rate of change per unit of time
of the amount of A in the control volume V is due to:
a) Generation of the property per unit mas and time time due to a source: k A ( x, t )
 
b) The convective (net incoming) flux across the surface of the volume. source term

c) The non-convective (net incoming) flux across the surface of the volume: jA (x, t )
 
 So, the global form of the general balance equation is: non-convective
flux vector

∂t V∫ ∫ ρ kA
ρ ψ=
dV dV − ∫ ρ ψ v ⋅ n dS − ∫ jA ⋅ n dS
 
V ∂V
  ∂V
a b c

31


∂t V
ρ ψ=
dV ∫ ρ kA dV − ∫ρψ v ⋅ n dS − ∫ jA ⋅ n dS
General Balance Equation
V ∂V ∂V

 The global form is rewritten using the Divergence Theorem and


the definition of local derivative:


∂t V
ρ ψ dV + ∫ ρ ψ v ⋅ n dS =
∂V

∂ 
= ∫V  ∂t ( ρ ψ ) + ∇ ⋅ ( ρ ψ v ) = ∫ ( ρ kA − ∇ ⋅ jA ) dV
 dV
V

=ρ (Reynolds Theorem)
dt

∫ ρ= dV ∫ ( ρ kA − ∇ ⋅ jA ) dV ∀∆V ⊂ V ∀t
∆V ⊂V
dt ∆V ⊂V

 The local spatial form of the general balance equation is:


REMARK dψ
dψ ρ = ρ kA − ∇ ⋅ jA
For only convective transport ( jA = 0) then ρ dt = ρ kA dt
and the variation of the contents of in a given particle
is only due to the internal generation ρ kA .

32

ρ = ρ kA − ∇ ⋅ jA
Example
dt
∂ dψ
( ρψ
= ) ρ − ∇ ⋅ ( ρ ψ v ) ∀x ∈V
∂t dt

If the property A is associated to mass A ≡ M , then:


 The amount of the property per unit of mass is ψ = 1.
 The mass generation source term is kM = 0.
 The mass conservation principle states that mass cannot be generated.
 The non-convective flux vector is jM = 0 .
 Mass cannot be transported in a non-convective form.

ρ = ρ kA − ∇ ⋅ jA= 0
dt  
=0 =0
Then, the local spatial form of the general balance equation is:
dψ ∂ ∂ρ
=
ρ ( ρ ψ
 ) + ∇ ⋅ ( ρ ψ
=
 v ) 0 + ∇ ⋅ ( ρ v) =
0
dt ∂t ∂t
= 1= 1
∂ρ dρ Two equivalent forms of
+ ∇ ⋅ ( ρ v )= + ρ ∇ ⋅ v = 0 ∀x ∈ V ∀t
∂t dt the continuity equation.

33
5.7. Linear Momentum Balance
Ch.5. Balance Principles

34
Linear Momentum
in Classical Mechanics
 Applying Newton’s 2nd Law to the discrete system formed by n
particles, the resulting force acting on the system is:
n n n
dv i
R ( t )= ∑
= fi ∑ m=
a ∑ m =
 =i 1 =i 1 i = i
i 1
i
dt
Resulting force mass conservation
on the system
principle: dmi = 0
d n n
dmi dP ( t ) dt
=
=
∑ mi v i − ∑
dt i 1 =i 1 dt
vi =
dt
= P ( t ) linear momentum
 For a system in equilibrium, = R 0, ∀t :
dP ( t )
=0 P ( t ) = cnt CONSERVATION OF THE
dt LINEAR MOMENTUM

35
Linear Momentum n
P ( t ) = ∑ mi v i
in Continuum Mechanics
i =1

 The linear momentum of a material volume Vt of a continuum


medium with mass M is:

=P (t ) v ( x, t ) d M ∫ ρ ( x, t ) v ( x, t ) dV
∫=
M V

d M = ρ dV

36
Linear Momentum Balance Principle
 The time-variation of the linear momentum of a material volume is
equal to the resultant force acting on the material volume.
dP ( t ) d
= = ∫ ρ v dV R ( t )
dt dt Vt

 Where: body forces


R (t )
= ∫ ρ b dV + ∫ t dS
V ∂V

surface forces

 If the body is in equilibrium, the linear momentum is conserved:


dP ( t )
R ( t ) = 0= 0= P ( t ) cnt
dt
37
Global Form of the
Linear Momentum Balance Principle
 The global form of the linear momentum balance principle:
d dP ( t )
R (t ) ∫ ρ b dV + =∫ t dS ∫ =ρ v dV ∀∆V ⊂ V , ∀t
∆V ⊂V ∂∆V ⊂V
dt ∆Vt ⊂Vt ≡V dt
 
P (t )

 Introducing t= n ⋅ σ and using the Divergence Theorem,

∫ t dS =∫ n ⋅ σ dS =∫ ∇ ⋅ σ dV
∂V ∂V V
 So, the global form is rewritten:
∫ ρ b dV + ∫ t dS =
∆V ⊂V ∂∆V ⊂V

d
= ∫ ( ρ b=
+∇ ⋅ σ ) dV ∫ ρ v dV ∀∆V ⊂ V , ∀t
38 ∆V ⊂V
dt ∆Vt ⊂Vt ≡V
Local Form of the
Linear Momentum Balance Principle
 Applying Reynolds Lemma to the global form of the principle:
d dv
∫ ( ∇=
⋅ σ + ρ b ) dV = ∫ ρ v dV ∫ ρ dV ∀∆V ⊂ V , ∀t
∆V ⊂V
dt ∆Vt ⊂Vt ≡V ∆V ⊂V
dt

 Localizing, the linear momentum balance principle reads:

∆V → dV (x, t )
dv(x, t )
∇ ⋅ σ (x, t ) + ρ=
b(x, t ) ρ = ρ a(x, t ) ∀x ∈ V , ∀t
dt
LOCAL FORM OF THE LINEAR
MOMENTUM BALANCE
(CAUCHY’S EQUATION OF MOTION)

39
5.8. Angular Momentum Balance
Ch.5. Balance Principles

40
Angular Momentum
in Classical Mechanics
 Applying Newton’s 2nd Law to the discrete system formed by n
particles, the resulting torque acting on the system is:
n n
dv i
) ∑ ri × f=i ∑ ri × mi =
MO ( t=
=i 1 =i 1 dt
=0 n
d n n
dri d dL
=
dt

=i 1 =i 1
ri × m v
i i − ∑ dt
× m v=
i i
=i 1 dt
∑ ri × m v=
i i
dt
= L (t )
= vi
angular momentum
d L (t )
MO (t ) =
dt
 For a system in equilibrium, M = O 0, ∀t :
d L (t )
= 0 ∀t L ( t ) = cnt CONSERVATION OF THE
dt ANGULAR MOMENTUM

41
Angular Momentum
in Continuum Mechanics
 The angular momentum of a material volume Vt of a continuum
medium with mass M is:

L (t ) =
∫ r × v ( x, t ) d M =
∫ x × ρ ( x, t ) v ( x, t ) dV
M ≡x V

d M = ρ dV

42
Angular Momentum Balance Principle
 The time-variation of the angular momentum of a material volume
with respect to a fixed point is equal to the resultant moment with
respect to this fixed point.

d L (t ) d
= ∫ r × ρ v dV = M O ( t )
dt dt Vt ≡V ≡x

 Where: torque due to


body forces

M O ( t ) =∫ r × ρ b dV + ∫ r × t dS
V ∂V
torque due to
surface forces

43
Global Form of the
Angular Momentum Balance Principle
 The global form of the angular momentum balance principle:
d
∫V (r × ρ b) dV + ∂∫V (r × t) dS
= ∫
dt Vt ≡V
(r × ρ v ) dV

 Introducing t= n ⋅ σ and using the Divergence Theorem,

∫ r × t dS= ∫ r × n ⋅ σ dS= ∫ × σ ⋅ n dS= ∫ ( r × σ ) ⋅ n dS=


T T
r
∂V ∂V ∂V ∂V

= ∫ ( r × σT ) ⋅ ∇ dV
V

It can be proven that,


 REMARK
( r × σT ) ⋅ ∇ = r × ∇ ⋅ σ + m ijk is the Levi-Civita
permutation symbol.

= m m= ˆ
i ei ; mi ijkσ jk

44
Global Form of the
Angular Momentum Balance Principle
 Applying Reynolds Lemma to the right-hand term of the global
form equation:
Reynold's
Lemma
d d ↓ d
∫ r × ρ v dV
= ∫ ρ ( r × v ) dV = ∫ρ ( r × v ) dV
=
dt Vt ≡V dt Vt ≡V V
dt
 dr =0 dv  dv
= ∫ ρ  × v + r ×  dV = ∫V r × ρ dt dV
 dt dt 
V
=v
 Then, the global form of the balance principle is rewritten:
dv
∫ r × ( ρ b + ∇ ⋅σ ) + ijkσ jk eˆ i  dV=
V
∫ r×ρ
V
dt
dV

45
Local Form of the
Angular Momentum Balance Principle
 Rearranging the equation:
=0 (Cauchy’s Eq.)
  dv  
∫  
∆V ⊂V 
r ×

∇ ⋅ σ + ρ b − ρ
dt


+ m = 0
 dV


∆V ⊂V
= 0 ∀∆V ⊂ V , ∀t
m( x, t )dV

 Localizing
m(x=
, t) 0 mi ijkσ=
= jk 0 ; i, j , k ∈ {1, 2,3} ; ∀x ∈ Vt , ∀t
= i 1 ⇒ 123 σ 23 + 132 σ 32= 0 ⇒ σ 23= σ 32
  σ 11 σ 12 σ 13 
 =1 =−1

i 2 ⇒ 231 σ 31 + 213 σ 13= 0 ⇒ σ 31= σ 13
= σ ≡ σ 12 σ 22 σ 23 
  
 =1 =−1 σ 13 σ 23 σ 33 
=i 3 ⇒ 312 σ 12 + 321 σ 21= 0 ⇒ σ 12= σ 21
  
=1 =−1
= σ ( x, t ) σ T ( x, t ) ∀x ∈ Vt , ∀t
SYMMETRY OF THE CAUCHY’S STRESS TENSOR
46
5.9. Mechanical Energy Balance
Ch.5. Balance Principles

47
Power
 Power, W (t ) , is the work performed in the system per unit of
time.

 In some cases, the power is an exact time-differential of a


function (then termed) energy E :
d E (t )
W (t ) =
dt
 It will be assumed that the continuous medium absorbs power
from the exterior through:
 Mechanical Power: the work performed by the mechanical actions
(body and surface forces) acting on the medium.
 Thermal Power: the heat entering the medium.

48
External Mechanical Power
 The external mechanical power is the work done by the body
forces and surface forces per unit of time.
 In spatial form it is defined as:
Pe ( t ) = ∫ ρ b ⋅ v dV + ∫ t ⋅ v dS
V ∂V

dr
ρ b ⋅ dV =⋅ ρ b v dV

dt
=v
dr
t ⋅ dS =t ⋅ v dS

dt
=v
Mechanical Energy Balance
 Using t= n ⋅ σ and the Divergence Theorem, the traction
contribution reads,
Divergence
Theorem

∫∂V
=
t ⋅ v=
dS ∫∂V n ⋅ (σ ⋅ v ) dS V
= ∫ ( ∇ ⋅ σ ) ⋅ v + σ : ∇v  dV
∫ ∇ ⋅ (σ ⋅ v ) dV V
n⋅σ =l
 Taking into account the identity: l= d + 
w spatial velocity
=0 skew gradient tensor
σ=
:l σ :d +σ :w symmetric

 So, ∫
∂V
t ⋅ v dS= ∫ ( ∇ ⋅ σ ) ⋅ v dV + ∫ σ : d dV
V V

50
dv
∇ ⋅σ + ρ b =
ρ
Mechanical Energy Balance dt

 Substituting and collecting terms, the external mechanical power


in spatial form is, ∫ t⋅v dS
 ∂ V
Pe ( t ) = ∫ ρ b ⋅ v dV + ∫ ( ∇ ⋅ σ ) ⋅ v dV + ∫ σ : d dV =
V V V

dv
=∫ ( ∇ ⋅ σ + ρ b ) ⋅ v dV + ∫ σ : d dV =∫ ρ ⋅ v dV + ∫ σ : d dV
V     dt 
V V V

= ρ dv d 1 d 1
dt = ρ ( v= ⋅ v ) ρ ( v )
2

dt 2 dt 2 v = v

Reynold's
Lemma
d 1 2 ↓ d 1 2
Pe ( t=) ∫ ρ ( v )dV + ∫ σ : d dV = ∫ ρ ( v )dV + ∫ σ : d dV
V
dt 2 V
dt V 2 V

51
Mechanical Energy Balance. Theorem of
the expended power. Stress power

d 1 2
Pe ( t ) = ∫ ρ b ⋅ v dV + ∫ t ⋅ v dS = ∫ ρ v dV + ∫ σ : d dV
V ∂V dt Vt ≡V 2 V

external mechanical power K


kinetic energy

entering the medium stress power

d
Pe ( t )
= K ( t ) + Pσ Theorem of the expended
dt mechanical power

REMARK
The stress power is the mechanical power entering the system which is not spent
in changing the kinetic energy. It can be interpreted as the work by unit of time
done by the stress in the deformation process of the medium.
A rigid solid will produce zero stress power ( d = 0 ) .

52
External Thermal Power
 The external thermal power is incoming heat in the continuum
medium per unit of time.
 The incoming heat can be due to:
 Non-convective heat transfer across the
volume’s surface.
incoming heat
− ∫ q(x, t ) ⋅ n dS =
 unit of time
∂Vheat conduction
flux vector

 Internal heat sources


heat generated by internal sources
∫V 
ρ r ( x , t ) dV =
unit of time
specific
internal heat
production

53
External Thermal Power
 The external thermal power is incoming heat in the continuum medium
per unit of time.
 In spatial form it is defined as:
Qe (=
t) ∫ ρ r dV − ∫ q ⋅ n dS
= ∫ ( ρ r − ∇ ⋅ q) dV
V

∂V
 V

= ∫∂V n⋅q dS
= ∫V (∇⋅q) dV
where:
q ( x, t ) is the non-convective heat flux vector per unit of spatial surface
r ( x, t ) is the internal heat source rate per unit of mass.

54
Total Power
 The total power entering the continuous medium is:

d 1 2
=
Pe + Qe ∫
dt Vt ≡V 2
ρ v dV + ∫ σ : d dV + ∫ ρ r dV − ∫ q ⋅ n dS
V V ∂V

55
5.10. Energy Balance
Ch.5. Balance Principles

56
Thermodynamic Concepts
 A thermodynamic system is a macroscopic region of the continuous
medium, always formed by the same collection of continuous matter
(material volume). It can be:
ISOLATED SYSTEM OPEN SYSTEM Thermodynamic space

MATTER

HEAT

 A thermodynamic system is characterized and defined by a set of


thermodynamic variables µ1, µ2, ....µn which define the thermodynamic
space.
 The set of thermodynamic variables necessary to uniquely define a
system is called the thermodynamic state of a system.

57
Thermodynamic Concepts
 A thermodynamic process is the energetic development of a
thermodynamic system which undergoes successive thermodynamic states,
changing from an initial state to a final state
→ Trajectory in the thermodynamic space.
 If the final state coincides with the initial state, it is a closed cycle process.

 A state function is a scalar, vector or tensor entity defined univocally as a


function of the thermodynamic variables for a given system.
 It is a property whose value does not depend on the path taken to reach that
specific value.
58
State Function
Is a function φ ( µ1 ,..., µn ) uniquely valued in terms of the “thermodynamic state”
or, equivalently, in terms of the thermodynamic variables {µ1 , µ2 , , µn }
 Consider a function φ ( µ1 , µ 2 ) , that is not a state function, implicitly defined in
the thermodynamic space by the differential form:
= δφ f1 ( µ1 , µ2 ) d µ1 + f 2 ( µ1 , µ2 ) d µ2
 The thermodynamic processes Γ and Γ 2 yield:
1
φ = φ A + ∫ δφ =
 B Γ1 ∫Γ1 f 2 (µ1 , µ2 )δµ2
 ∫Γ1 δφ ≠ ∫Γ2 δφ φB' ≠ φB
φ A ∫ δφ =
φB ' =+ ∫ f 2 ( µ1 , µ2 )δµ2
 Γ 2 Γ 2

 For to be a state function, the differential form must be


an exact differential: , i.e., must be integrable:
 The necessary and sufficient condition for this is the equality of cross-derivatives:
∂fi ( µ1 ,..., µn ) ∂f j ( µ1 ,..., µn )
= ∀i, j ∈ {1,...n} δφ = dφ
∂µ j ∂µi
59
First Law of Thermodynamics
POSTULATES:
1. There exists a state function E ( t ) named total energy of the system, such that its
material time derivative is equal to the total power entering the system:
d d 1 2
E ( t ) := Pe ( t ) + Qe ( t )=
dt Vt∫≡V 2
ρ v dV + ∫ σ : d dV + ∫ ρ r dV − ∫ q ⋅ n dS
dt
  V
 V ∂V

Pe (t ) Qe (t )

2. There exists a function U ( t ) named the internal energy of the system, such that:
 It is an extensive property, so it can be defined in terms of a specific internal energy (or
internal energy per unit of mass) u ( x, t ):
U ( t ) := ∫ ρ u dV REMARK
d E and d Kare exact differentials,
V
 The variation of the total energy of the system is:
U d E − d K.
therefore, so is d=
d d d Then, the internal energy is a
= E (t ) K (t ) + U (t )
dt dt dt state function.
60
Global Form of the
Internal Energy Balance
 Introducing the expression for the total power into the first
postulate: =K
d d 1 2
= E (t ) ∫ ρ v dV + ∫ σ : d dV + ∫ ρ r dV − ∫ q ⋅ n dS
dt dt Vt ≡V 2 V V ∂V

 Comparing this to the expression in the second postulate:


d d d
= E (t ) K (t ) + U (t )
dt dt dt

 The internal energy of the system must be:


d d GLOBAL FORM
(t )
U= ∫ ρ u=
dV ∫ σ : d dV + ∫ ρ r dV − ∫ q ⋅ n dS OF THE INTERNAL
dt dt Vt ≡V V V ∂V
ENERGY BALANCE
Pσ ( t ) Q, e ( t )
stress power external thermal power
61
Local Spatial Form of the
Internal Energy Balance
 Applying Reynolds Lemma to the global form of the balance
equation, and using the Divergence Theorem:
d d du
(t )
dt ∆Vt ⊂∫Vt ≡V ∫ ∫ ∫ ∫
U= ρ u=
dV ρ =dV σ : d dV + ρ r dV − q ⋅ n dS
dt dt
  ∆Vt ⊂Vt ≡V ∆V ⊂V ∆V ⊂V
 
∂∆V ⊂V

U (t ) ∫
∂∆V ⊂V
∇⋅q dV

du
⇒ =∫
∆V ⊂V
ρ
dt
dV ∫
∆V ⊂V
σ : d dV + ∫
∆V ⊂V
ρ r dV − ∫
∆V ⊂V
∇ ⋅ q dV ∀∆V ⊂ V ∀t

 Then, the local spatial form of the energy balance principle is


obtained through localization ∆V → dV (x, t ) as:
du LOCAL FORM OF THE
ρ= σ : d + ( ρ r − ∇ ⋅ q ) ∀x ∈ V , ∀t ENERGY BALANCE
dt (Energy equation)

62
Second Law of Thermodynamics
 The total energy is balanced in all thermodynamics processes
following:
dE dK dU
Pe ( t ) + Qe ( t ) = = +
dt dt dt
 In an isolated system (no work can enter or exit the system)
dE dU dK
Pe ( t ) + Qe ( t ) = =0 + =
0
dt dt dt

 However, it is not established if the energy exchange can happen


in both senses or not:
dU dK dU dK
>0 <0 <0 >0
dt dt dt dt
 There is no restriction indicating if an imagined arbitrary process is
physically possible or not.

63
Second Law of Thermodynamics
 The concept of energy in the first law does not account for
the observation that natural processes have a preferred
direction of progress. For example:

 If a brake is applied on a spinning wheel, the


speed is reduced due to the conversion of kinetic
energy into heat (internal energy). This process
never occurs the other way round. dU dK
>0 <0
dt dt
 Spontaneously, heat always flows to regions of
lower temperature, never to regions of higher
temperature.

64 MMC - ETSECCPB - UPC 24/04/2017


Reversible and Irreversible Processes
 A reversible process can be “reversed” by means of infinitesimal
changes in some property of the system.
 It is possible to return from the final state to the initial state along the same path.
 A process that is not reversible is termed irreversible.
REVERSIBLE PROCESS IRREVERSIBLE PROCESS

 The second law of thermodynamics allows discriminating:


IMPOSSIBLE
thermodynamic processes REVERSIBLE
POSSIBLE
IRREVERSIBLE
65
Second Law of Thermodynamics
POSTULATES:
1. There exists a state function θ ( x,t ) denoted absolute temperature,
which is always positive.

2. There exists a state function S named entropy, such that:


 It is an extensive property, so it can be defined in terms of a specific entropy
or entropy per unit of mass s :
S (t ) = ∫ ρ s (x, t )dV
V
 The following inequality holds true:
d d r q Global form of the 2nd
=
dt
S (t ) ∫
dt V
ρ s dV ≥ ∫ ρ dV − ∫ ⋅ n dS
θ θ
Law of
V ∂V Thermodynamics
= reversible process
> irreversible process
66
Second Law of Thermodynamics
SECOND LAW OF THERMODYNAMICS IN CONTINUUM MECHANICS
The rate of the total entropy of the system is equal o greater than the rate of heat per
unit of temperature

d d r q Global form of the 2nd


=
dt
S (t ) ∫
dt V
ρ s dV ≥ ∫ ρ dV − ∫ ⋅ n dS
θ θ
Law of
V ∂V
Thermodynamics
= reversible process = Γe ( t )
> irreversible process
rate of the total amount of the entity heat, per unit
(t )
Qe= ∫ ρ r dV − ∫ q ⋅ n dS of time, (external thermal power) entering into the
V ∂V system

r q rate of the total amount of the entity heat per unit


e (t )
Γ= ∫ρ dV − ∫θ ⋅ n dS of absolute temperature, per unit of time (external
V
θ ∂V heat/unit of temperature power) entering into the
system
67
Second Law of Thermodynamics

 Consider the decomposition of entropy into two (extensive) counterparts:


 Entropy generated inside the continuous medium:

S (t ) S ( ) (t ) + S ( (t )
e)
=
i

dS dS ( ) dS ( )
i e
= +
dt dt dt
 Entropy generated by interaction with the outside medium:

S ( ) = ∫ ρ s( ) ( x, t ) dV
i i

S ( e ) = ∫ ρ s( e ) ( x, t ) dV
V

68
Second Law of Thermodynamics
dS ( )
e
If one establishes, r q
=Γ e =∫ ρ dV − ∫ ⋅ n dS

dt V
θ ∂V
θ

 Then the following must hold true:


dS (i ) dS ( e ) dS r q
+ = ≥ ∫ ρ dV − ∫ ⋅ n dS
dt dt dt V θ θ
 ∂V

(e)
 And thus, = dS
dt

dS (i ) dS dS ( e ) dS  r q 
= − = −  ∫ ρ dV − ∫ ⋅ n dS  ≥ 0 ∀∆V ⊂ V ∀t
dt dt dt dt  ∆V ⊂V θ ∂∆V ⊂V
θ 

REPHRASED SECOND LAW OF THERMODYNAMICS :


()
i
The internally generated entropy of the system , S (t ) , never decreases along time

69
Local Spatial Form of the
Second Law of Thermodynamics
 The previous eq. can be rewritten as:
d (i ) d  r q 
= ∫
dt ∆Vt ⊂Vt ≡V
ρ s dV ∫
dt ∆Vt ⊂Vt ≡V
ρ s dV −  ∫ ρ
 ∆V ⊂V θ
dV − ∫
∂∆V ⊂V
θ
⋅ n dS ≥0

∀∆V ⊂ V ∀t

 Applying the Reynolds Lemma and the Divergence Theorem:

ds (i ) ds  r q 
∫ =
∆V ⊂V
ρ
dt
dV ∫
∆V ⊂V
ρ
dt
dV −  ∫ ρ dV − ∫ ∇ ⋅   dV  ≥ 0 ∀∆V ⊂ V ∀t
 ∆V ⊂V θ ∆V ⊂V θ  

 Then, the local spatial form of the second law of thermodynamics is:
ds (i ) ds  r  q  Local (spatial) form of the 2nd
ρ = ρ −  ρ − ∇ ⋅  ≥ 0 ∀x ∈ V , ∀t Law of Thermodynamics
dt dt  θ θ  (Clausius-Duhem inequality)

= reversible process
> irreversible process

70
Local Spatial Form of the
Second Law of Thermodynamics
Considering that, q 1 1 REMARK
 ∇ ⋅ ( )= ∇ ⋅ q − 2 q ⋅ ∇θ
θ θ θ (Stronger postulate)
Internally generated entropy can
 The Clausius-Duhem inequality can be written as (i )
be generated locally, slocal , or by
= s( ) = s
i
(i )
thermal conduction, scond , and
ds (i )  ds r 1  1 both must be non-negative.
=  − + ∇ ⋅q  − q ⋅ ∇θ ≥ 0
dt  dt θ ρθ  ρθ 2

(i ) (i )
= slocal = scond
CLAUSIUS-PLANCK HEAT FLOW
 r 1 
 
s − + ∇ ⋅ q  ≥0 −
1
q ⋅ ∇θ ≥ 0 INEQUALITY
 θ ρθ  INEQUALITY ρθ
2

Because density and absolute temperature are


always positive, it is deduced that q ⋅ ∇θ ≤ 0 ,
which is the mathematical expression for the fact
that heat flows by conduction from the hot parts of
the medium to the cold ones.
71
Alternative Forms of the
Clausius-Planck Inequality
 Substituting the internal energy balance equation given by
du not
ρ = ρ= u σ : d + ρ r − ∇ ⋅ q ρ r σ : d − ρ u
∇ ⋅q −=
dt

into the Clausius-Planck inequality,

ρθ silocal
=: ρθ s − ρ r + ∇ ⋅ q ≥ 0

yields,
ρθ s + (σ : d − ρ u ) ≥ 0 − ρ ( u − θ s ) + σ : d ≥ 0

Clausius-Planck Inequality
in terms of the
specific internal energy

72
5.11. Governing Equations
Ch.5. Balance Principles

74
Governing Equations in Spatial Form

Conservation of Mass.
ρ + ρ ∇ ⋅ v = 0 Continuity Equation.
1 eqn.

Linear Momentum Balance.


∇ ⋅ σ + ρ b = ρ v 3 eqns.
First Cauchy’s Motion Equation.

σ =σT Angular Momentum Balance.


3 eqns.
Symmetry of Cauchy Stress Tensor.

ρ u σ : d + ρ r − ∇ ⋅ q
= Energy Balance.
1 eqn.
First Law of Thermodynamics.

− ρ ( u − θ s ) + σ : d ≥ 0 Second Law of Thermodynamics.


1 Clausius-Planck Inequality. 2 restrictions
− 2 q ⋅ ∇θ ≥ 0 Heat flow inequality
ρθ
8 PDE +
2 restrictions

75
Governing Equations in Spatial Form
 The fundamental governing equations involve the following variables:
ρ density 1 variable
v velocity vector field 3 variables
σ Cauchy’s stress tensor field 9 variables
u specific internal energy 1 variable
q heat flux per unit of surface vector field 3 variables

θ absolute temperature 1 variable


19 scalar
s specific entropy 1 variable
unknowns

 At least 11 equations more (assuming they do not involve new unknowns),


are needed to solve the problem, plus a suitable set of boundary and
initial conditions.
76
Constitutive Equations in Spatial Form

σ = σ ( v, θ , ζ ) Thermo-Mechanical
6 eqns.
Constitutive Equations.

s = s ( v, θ , ζ ) Entropy
Constitutive Equation. 1 eqn.

Thermal Constitutive Equation.


q = q ( v, θ ) = − K ∇ θ Fourier’s Law of Conduction. 3 eqns.

u = f ( ρ , v, θ , ζ ) Heat
State Equations. (1+p) eqns.
Fi ( ρ ,θ=
, ζ ) 0 i ∈ {1, 2,..., p} Kinetic
(19+p) PDE +
set of new thermodynamic
{ }
variables:ζ = ζ 1 , ζ 2 ,..., ζ p .
(19+p) unknowns

REMARK 1 REMARK 2
The strain tensor is not considered an unknown as they These equations are
can be obtained through the motion equations, i.e., ε = ε ( v ). specific to each material.

77
The Coupled
Thermo-Mechanical Problem
Conservation of Mass.
ρ + ρ ∇ ⋅ v = 0 Continuity Mass Equation.
1 eqn.

16 scalar Linear Momentum Balance. 10


unknowns 3 eqns. equations
First Cauchy’s Motion Equation.

σ = σ ( ε ( v),θ ) Mechanical constitutive equations. 6 eqns.

Energy Balance.
1 eqn.
First Law of Thermodynamics.

Second Law of Thermodynamics.


2 restrictions.
Clausius-Planck Inequality.

78 MMC - ETSECCPB - UPC


The Uncoupled
Thermo-Mechanical Problem
 The mechanical and thermal problem can be uncoupled if
 The temperature distribution θ ( x,t ) is known a priori or does not intervene in
the mechanical constitutive equations.

 Then, the mechanical problem can be solved independently.

79
The Uncoupled
Thermo-Mechanical Problem
Conservation of Mass.
ρ + ρ ∇ ⋅ v = 0 Continuity Mass Equation.
1 eqn.

10 scalar Linear Momentum Balance. Mechanical


unknowns 3 eqns. problem
First Cauchy’s Motion Equation.

σ = σ ( ε ( v), θ ) Mechanical constitutive equations. 6 eqns.

Energy Balance.
1 eqn.
First Law of Thermodynamics.
Thermal
problem
Second Law of Thermodynamics.
2 restrictions.
Clausius-Planck Inequality.

80
The Uncoupled
Thermo-Mechanical Problem
 Then, the variables involved in the mechanical problem are:
ρ density 1 variable
Mechanical v velocity vector field 3 variables
variables
σ Cauchy’s stress tensor field 6 variables
u specific internal energy 1 variable
q heat flux per unit of surface vector field 3 variables
Thermal
variables
θ absolute temperature 1 variable
s specific entropy 1 variable

81
Chapter 5
Balance Principles

rs
ee
s gin
5.1 Introduction

t d le En

r
Continuum Mechanics is based on a series of general postulates or principles

ba
ge ro or
eS m
that are assumed to always be valid, regardless of the type of material and the

ci
f
range of displacements or deformations. Among these are the so-called balance

ra
C d P cs
principles:
b
a
i
an an n

• Conservation of mass
y ha

• Balance of linear momentum


le
• Balance of angular momentum
liv or ec

• Balance of energy (or first law of thermodynamics)


M

.A

A restriction that cannot be rigorously understood as a balance principle must


m

be added to these laws, which is introduced by the


d
uu
e

• Second law of thermodynamics


X Th

er
tin

5.2 Mass Transport or Convective Flux


on

.O
C

In continuum mechanics, the term convection is associated with mass transport


©

in the medium, which derives from the motion of its particles. The continuous
medium is composed of particles, some of whose properties are associated with
the amount of mass: specific weight, angular momentum, kinetic energy, etc.
Then, when particles move and transport their mass, a transport of the these
properties occurs, named convective transport (see Figure 5.1).
Consider A, an arbitrary (scalar, vector or tensor) property of the continuous
medium, and Ψ (x,t), the description of the amount of said property per unit of
mass of the continuous medium. Consider also S, a control surface, i.e., a surface
fixed in space (see Figure 5.2). Due to the motion of the particles in the medium,
these cross the surface along time and, in consequence, there exists a certain
amount of the property A that, associated with the mass transport, crosses the
control surface S per unit of time.

193
194 C HAPTER 5. BALANCE P RINCIPLES

rs
ee
s gin
Figure 5.1: Convective transport in the continuous medium.

t d le En

r
ba
ge ro or
eS m
ci
f
Definition 5.1. The convective flux (or mass transport flux) of a

ra
C d P cs
generic property A through a control surface S is the amount of A
b
a
i
that, due to mass transport, crosses the surface S per unit of time.
an an n
y ha

convective flux not amount of A crossing S


= ΦS =
le
liv or ec

of A through S unit of time


M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 5.2: Convective flux through a control surface.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Mass Transport or Convective Flux 195

Figure 5.3: Cylinder occupied by the particles that have crossed dS in the time inter-

rs
val [t, t + dt].

ee
s gin
To obtain the mathematical expression of the convective flux of A through the

t d le En
surface S, consider a differential surface element dS and the velocity vector v of
the particles that at time t are on dS (see Figure 5.3). In a time differential dt,

r
these particles will have followed a pathline dx = v dt, such that at the instant of

ba
ge ro or
eS m
time t + dt they will occupy a new position in space. Taking now into account all

ci
f

ra
the particles that have crossed dS in the time interval [t, t + dt], these will occupy
C d P cs
b
a
a cylinder generated by translating the base dS along the directrix dx = v dt, and
i
an an n

whose volume is given by


y ha

dV = dS dh = v · n dt dS . (5.1)
le
liv or ec

Since the volume (dV ) of the particles crossing dS in the time interval
M

.A

[t, t + dt] is known, the mass crossing dS in this same time interval can be ob-
tained by multiplying (5.1) by the density,
m

dm = ρ dV = ρv · n dt dS . (5.2)
uu
e
X Th

er

Finally, the amount of A crossing dS in the time interval [t, t + dt] is calculated
tin

by multiplying (5.2) by the function Ψ (amount of A per unit of mass),


on

.O

Ψ dm = ρ Ψ v · n dt dS . (5.3)
C

Dividing (5.3) by dt yields the amount of the property that crosses the differ-
ential control surface dS per unit of time,
Ψ dm
d ΦS = = ρ Ψ v · n dS . (5.4)
dt
Integrating (5.4) over the control surface S results in the amount of the property
A crossing the whole surface S per unit of time, that is, the convective flux of the
property A through S.
 
convective flux
ΦS = ρ Ψ v · n dS (5.5)
of A through S
S

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
196 C HAPTER 5. BALANCE P RINCIPLES

Example 5.1 – Compute the magnitude Ψ and the convective flux ΦS corre-
sponding to the following properties: a) volume, b) mass, c) linear momen-
tum, d) kinetic energy.

Solution
a) If the property A is the volume occupied by the particles, then Ψ is the
volume per unit of mass, that is, the inverse of the density. Therefore,

1
A ≡V and Ψ = lead to ΦS = v · n dS = volume flow rate .
ρ

rs
S

ee
b) If the property A is the mass, then Ψ is the mass per unit of mass, that

s gin
is, the unit. Therefore,


t d le En
A ≡ M and Ψ = 1 lead to ΦS = ρ v · n dS .

r
ba
ge ro or
eS m
S

ci
f

ra
c) If the property A is the linear momentum (= mass × velocity), then Ψ
C d P cs
b
a
is the linear momentum per unit of mass, that is, the velocity. Therefore,
i
an an n


y ha

A ≡ m v and Ψ = v lead to ΦS = ρ v (v · n) dS .
le
liv or ec

S
(Note that in this case Ψ and the convective flux ΦS are vectors).
M

.A

d) If the property A is the kinetic energy then Ψ is the kinetic energy per
m

unit of mass. Therefore,


uu
e


X Th

er

1 1 1
tin

A ≡ m |v|2 and Ψ = |v|2 lead to ΦS = ρ |v|2 (v · n) dS .


2 2 2
on

.O

S
C

Remark 5.1. In a closed control surface1 , S = ∂V , the expression of


the convective flux corresponds to the net outflow, defined as the
outflow minus the inflow (see Figure 5.4), that is,

not
net convective flux of A = Φ∂V = ρ Ψ v · n dS .
∂V

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Mass Transport or Convective Flux 197

rs
ee
s gin
Figure 5.4: Net outflow through a closed control surface.

t d le En

r
ba
ge ro or
eS m
ci
f
Remark 5.2. The convective flux of any property through a material

ra
C d P cs
b
surface is always null. Indeed, the convective flux of any property is

a
i
associated, by definition, with the mass transport (of particles) and,
an an n
y ha

on the other hand, a material surface is always formed by the same


particles and cannot be crossed by them. Consequently, there is no
le
liv or ec

mass transport through a material surface and, therefore, there is no


M

.A

convective flux through it.


m

d
uu
e
X Th

er
tin

Remark 5.3. Some properties can be transported within a continuous


medium in a manner not necessarily associated with mass transport.
on

.O

This form of non-convective transport receives several names (con-


C

duction, diffusion, etc.) depending on the physical problem being


©

studied. A typical example is heat flux by conduction.


The non-convective transport of a property is characterized by the
non-convective flux vector (or tensor) q (x,t), which allows defining
the (non-convective) flux through a surface S with unit normal vector
n as 
non-convective flux = q · n dS .
S

1 Unless stated otherwise, when dealing with closed surfaces, the positive direction of the
unit normal vector n is taken in the outward direction of the surface.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
198 C HAPTER 5. BALANCE P RINCIPLES

5.3 Local and Material Derivatives of a Volume Integral


Consider A, an arbitrary (scalar, vector or tensor) property of the continuous
medium, and μ, the description of the amount of said property per unit of vol-
ume2 , amount of A
μ (x,t) = . (5.6)
unit of volume
Consider an arbitrary volume V in space. At time t, the total amount Q (t) of the
property contained in this volume is

Q (t) = μ (x,t) dV . (5.7)

rs
V

ee
To compute the content of property A at a different time t + Δt, the following

s gin
two situations arise:

t d le En
1) A control volume V is considered and, therefore, it is fixed in space and
crossed by the particles along time.

r
ba
ge ro or
eS m
2) A material volume that at time t occupies the spatial volume Vt ≡ V is

ci
f
considered and, thus, the volume occupies different positions in space

ra
C d P cs
b
along time.

a
i
an an n

Different values of the amount Q (t + Δt) are obtained for each case, and com-
y ha

puting the difference between the amounts Q (t + Δt) and Q (t) when Δt → 0
le
yields
liv or ec

Q (t + Δt) − Q (t)
Q (t) = lim , (5.8)
M

.A

Δt→0 Δt
resulting in two different definitions of the time derivative, which lead to the
m

concepts of local derivative and material derivative of a volume integral.


uu
e
X Th

er

5.3.1 Local Derivative


tin
on

.O

Definition 5.2. The local derivative



of the volume integral,
C

Q (t) = μ (x,t) dV ,
V
is the time derivative of Q (t) when the volume V is a volume fixed
in space (control volume), see Figure 5.5. The notation
not ∂ 
local derivative = μ (x,t) dV
∂t V
will be used.

2μ is related to Ψ = (amount of A)/(unit of mass) through μ = ρ Ψ and has the same tensor
order as the property A .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Local and Material Derivatives of a Volume Integral 199

rs
ee
Figure 5.5: Local derivative of a volume integral.

s gin
The amount Q of the generic property A in the control volume V at times t

t d le En
and t + Δt is, respectively,

r
 

ba
ge ro or
eS m
Q (t) = μ (x,t) dV and Q (t + Δt) = μ (x,t + Δt) dV . (5.9)

ci
f

ra
C d P cs
V V

b
a
i
Using (5.9) in addition to the concept of time derivative of Q (t) results in3
an an n
y ha


 ∂ 1 
le
Q (t) = μ (x,t) dV = lim Q (t + Δt) − Q (t) =
liv or ec

∂t Δt→0 Δt
⎛ ⎞
M

V
.A

 
1 ⎝
= lim μ (x,t + Δt) dV − μ (x,t) dV ⎠ =
m

Δt→0 Δt
d
uu

V V
e

 
X Th

er

μ (x,t + Δt) − μ (x,t) ∂ μ (x,t)


tin

= lim dV = dV ,
Δt→0 Δt ∂t
 
on

.O

V V
∂ μ (x,t) local
C

derivative
∂t of μ
©

(5.10)
which yields the mathematical expression of the local derivative of a volume
integral.

Local derivative of a volume integral


 
∂ ∂ μ (x,t) (5.11)
μ (x,t) dV = dV
∂t ∂t
V V

3 Note that the integration domain does not vary when the volume V is considered as a control
volume and, therefore, is fixed in space.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
200 C HAPTER 5. BALANCE P RINCIPLES

5.3.2 Material Derivative

Definition 5.3. The material derivative of the volume integral,



Q (t) = μ (x,t) dV ,
V
is the time derivative of Q (t) when the volume Vt is a material vol-
ume (mobile in space), see Figure 5.6. The notation
not d 
material derivative = μ (x,t) dV

rs
dt Vt

ee
will be used.

s gin
t d le En
The content Q of the generic property A in the material volume Vt at times t and

r
t+Δt is, respectively,

ba
ge ro or
eS m
 

ci
Q (t) = μ (x,t) dV f
Q (t + Δt) = μ (x,t + Δt) dV .

ra
and (5.12)
C d P cs
b
a
Vt Vt+Δt
i
an an n
y ha

Then, the material derivative is mathematically expressed as4


le
liv or ec


 
d  Q (t + Δt) − Q (t)
M

.A

Q (t) = μ (x,t) dV  = lim =


dt  Δt→0 Δt
m

V ≡V
Vt ⎛ t
⎞ (5.13)
d

 
uu
e

1 ⎝
= lim μ (x,t + Δt) dV − μ (x,t) dV ⎠ .
X Th

er

Δt→0 Δt
tin

Vt+Δt Vt
on

.O

The following step consists in introducing two variable substitutions, each


C

suitable for one of the two integrals in (5.13), which lead to the same integra-
©

tion domain in both expressions. These variable substitutions are given by the
equation of motion x = ϕ (X,t), particularized for times t and t + Δt,


⎪ xt = ϕ (X,t) → (dx1 dx2 dx3 )t = |F (X,t)| (dX1 dX2 dX3 ) ,

⎪  


dVt dV0

⎪ xt+Δt = ϕ (X,t + Δt) → (dx1 dx2 dx3 )t+Δt = |F (X,t + Δt)| (dX1 dX2 dX3 ) ,

⎪ 

⎩ 
dVt+Δt dV0
(5.14)
4 Note that the integration domains are now different at times t and t + Δt.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Local and Material Derivatives of a Volume Integral 201

rs
Figure 5.6: Material derivative of a volume integral.

ee
s gin
where the identity dVt = |F (X,t)| dV0 has been taken into account. The variable

t d le En
substitutions in (5.14) are introduced in (5.13), resulting in

r
 μ̄ (X,t + Δt)

ba
ge ro or
 

eS m
d 1

ci
μ (x,t) dV = lim μ (x (X,t + Δt) ,t + Δt) |F (X,t + Δt)| dV0
f

ra
dt Δt→0 Δt 
C d P cs
b
a
Vt V0
− μ (x (X,t) ,t) |F (X,t)| dV0 =
i

an an n
y ha

V0
μ̄ (X,t)
le

liv or ec

μ̄ (X,t + Δt) |F (X,t + Δt)| − μ̄ (X,t) |F (X,t)|


= lim dV0 =
M

.A

Δt→0 Δt
V0 
∂  d 
m

μ̄ (X,t) |F (X,t)| = μ (x,t) |F (x,t)|


uu

∂t
e

dt

d 
X Th

er
tin

= μ |F| dV0 .
dt
on

.O

V0 (5.15)
Finally, expanding the last integral in (5.15) 5 and considering the equality
C

d |F|/dt = |F| ∇ · v yields


©

  
 
d d  dμ d |F|
μ (x,t) dV = μ |F| dV0 = |F| + μ dV0 =
dt dt dt dt
Vt V0 V0 
|F| ∇ · v
     
dμ dμ
= + μ∇ · v |F| dV0 = + μ∇ · v dV ,
dt  dt
V0 dVt Vt
(5.16)
5 The change of variable xt = ϕ (X,t) is undone here.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
202 C HAPTER 5. BALANCE P RINCIPLES

that is6 ,

     
d  d dμ
μ (x,t) dV 
not
= μ (x,t) dV = + μ∇ · v dV . (5.17)
dt  dt dt
Vt Vt ≡V Vt ≡V V

Recalling the expression of the material derivative of a property (1.15) results in


 
 
d ∂μ
μ (x,t) dV = + v · ∇μ + μ∇ · v dV =
∂t

rs
dt
Vt ≡V V 

ee
∇ · (μv) (5.18)
   

s gin
∂μ ∂
= dV + ∇ · (μv) dV = μ dV + ∇ · (μv) dV ,
∂t ∂t

t d le En
V V V V

r
where the expression of the local derivative (5.11) has been taken into account.

ba
ge ro or
eS m
Then, (5.18) produces the expression of the material derivative of a volume in-

ci
f

ra
tegral.
C d P cs
b
a
i
an an n

Material derivative of a volume integral


  
y ha

d ∂
μ (x,t) dV = μ dV + ∇ · (μv) dV
le
liv or ec

dt ∂t (5.19)
Vt ≡V V V
  
M

.A

material local convective


derivative derivative
m

derivative
d
uu
e
X Th

er
tin

Remark 5.4. The form of the material derivative, given as a sum of a


local derivative and a convective derivative, that appears when differ-
on

.O

entiating properties of the continuous medium (see Chapter 1, Sec-


C

tion 1.4) also appears here when differentiating integrals in the con-
©

tinuous medium. Again, the convective derivative is associated with


the existence of a velocity (or motion) in the medium and, thus, with
the possibility of mass transport.

6 The expression 
d
μ (x,t) dV
dt
Vt ≡V
denotes the time derivative of the integral over the material volume Vt (material derivative of
the volume integral) particularized at time t, when the material volume occupies the spatial
volume V .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Conservation of Mass. Mass continuity Equation 203

rs
ee
Figure 5.7: Principle of conservation of mass in a continuous medium.

s gin
5.4 Conservation of Mass. Mass continuity Equation

t d le En

r
ba
ge ro or
eS m
ci
f

ra
Definition 5.4. Principle of conservation of mass. The mass of a
C d P cs
b
a
continuous medium (and, therefore, the mass of any material vol-
i
an an n

ume belonging to this medium) is always the same.


y ha

le
liv or ec

Consider a material volume Vt that at times t and t + Δt occupies the volumes


M

.A

in space Vt and Vt+Δt , respectively (see Figure 5.7). Consider also the spatial
m

description of the density, ρ (x,t). The mass enclosed by the material volume V
d

at times t and t + Δt is, respectively,


uu
e
X Th

 
er
tin

M (t) = ρ (x,t) dV and M (t + Δt) = ρ (x,t + Δt) dV . (5.20)


on

.O

Vt Vt+Δt
C

By virtue of the principle of conservation of mass, M (t) = M (t + Δt) must be


©

satisfied.

5.4.1 Spatial Form of the Principle of Conservation of Mass. Mass


Continuity Equation
The mathematical expression of the principle of conservation of mass of the
material volume M (t) is that the material derivative of the integral (5.20) is
null, 
 d
M (t) = ρ dV = 0 ∀t . (5.21)
dt
Vt

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
204 C HAPTER 5. BALANCE P RINCIPLES

By means of the expression of the material derivative of a volume integral (5.17),


the integral (or global) spatial form of the principle of conservation of mass
results in

Global spatial form of the principle of conservation of mass


   
d dρ , (5.22)
ρ dV = + ρ ∇ · v dV = 0 ∀ΔVt ⊂ Vt , ∀t
dt dt
Vt Vt
(ΔVt ) (ΔVt )

rs
which must be satisfied for Vt and, also, for any partial material volume ΔVt ⊂ Vt

ee
that could be considered. In particular, it must be satisfied for each of the ele-

s gin
mental material volumes associated with the different particles in the continuous
medium that occupy the differential volumes dVt . Applying (5.22) on each dif-

t d le En
ferential volume dVt ≡ dV (x,t) yields7

r
    

ba
ge ro or
eS m
dρ dρ (x,t)

ci
+ ρ∇ · v dV = + ρ (x,t) ∇ · v (x,t) dV (x,t) = 0
f

ra
dt dt
C d P cs
dV (x,t)
b
a
∀x ∈ Vt , ∀t
i
an an n


y ha

=⇒ + ρ∇ · v = 0 dV ∀x ∈ Vt , ∀t
dt
le
liv or ec

(5.23)
M

.A

Local spatial form of the principle of conservation of mass


(mass continuity equation)
m

(5.24)
d
uu


e

+ ρ∇ · v = 0 dV ∀x ∈ Vt , ∀t
X Th

dt
er
tin

which constitutes the so-called mass continuity equation. Replacing the expres-
on

.O

sion of the material derivative of the spatial description of a property (1.15) in


C

(5.24) results in
©

∂ρ ∂ρ
+ v · ∇ρ + ρ∇ · v = 0 =⇒ + ∇ · (ρv) = 0 , (5.25)
∂t  ∂t
∇ · (ρv)

which yields an alternative expression of the mass continuity equation.

7 This procedure, which allows reducing a global (or integral) expression such as (5.22) to
a local (or differential) one such as (5.24), is named in continuum mechanics localization
process.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Conservation of Mass. Mass continuity Equation 205


∂ρ ⎪



+ ∇ · (ρv) = 0 ⎪

∂t ⎪



∂ ρ ∂ (ρvi ) ∀x ∈ Vt , ∀t (5.26)
+ =0 i ∈ {1, 2, 3}

∂t ∂ xi ⎪





∂ ρ ∂ (ρvx ) ∂ (ρvy ) ∂ (ρvz ) ⎪
+ + + = 0⎪

∂t ∂x ∂y ∂z

rs
5.4.2 Material Form of the Principle of Conservation of Mass

ee
From (5.22)8 ,

s gin
     
dρ dρ 1 d |F|

t d le En
+ ρ∇ · v dV = +ρ dV =
dt dt |F| dt

r
Vt
  Vt  
1 d 

ba
ge ro or
d |F|

eS m
1 dρ
= |F| +ρ dV = ρ |F| dV =

ci
|F| dt dt f |F| dt 

ra
C d P cs
Vt  Vt
|F| dV0
b
a
d 
i
an an n

ρ |F|
y ha

  dt 

le
= ρ (X,t) |F (X,t)| dV0 ∀ΔV0 ⊂ V0 , ∀t ,
liv or ec

∂t
M

.A

V0
(5.27)
m

where the integration domain is now the volume in the reference configura-
d

tion, V0 . Given that (5.27) must be satisfied for each and every part ΔV0 of V0 , a
uu
e

localization process can be applied, which results in9


X Th

er
tin

∂ 
ρ (X,t) |F (X,t)| = 0 ∀X ∈ V0 , ∀t
on

.O

∂t
C

=⇒ ρ (X,t) |F (X,t)| = ρ (X) |F (X)| ∀t


©

=⇒ ρ (X, 0) |F| (X, 0) = ρ (X,t) |F| (X,t) =⇒ ρ0 |F|0 = ρt |F|t .


  
not not =1
= ρ0 |F|0 = ρt |F|t
(5.28)
Local material form of the mass conservation principle
(5.29)
ρ0 (X) = ρt (X) |F|t (X) ∀X ∈ V0 , ∀t

8 Here, the expression deduced in Chapter 2, d |F| / dt = |F| · ∇ · v , is considered.


9 The equality F (X, 0) = 1 =⇒ |F|0 = 1 is used here.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
206 C HAPTER 5. BALANCE P RINCIPLES

5.5 Balance Equation. Reynolds Transport Theorem


Consider A, an arbitrary (scalar, vector or tensor) property of the continuous
medium, and Ψ (x,t), the description of the amount of said property per unit of
mass. Then, ρΨ (x,t) is the amount of this property per unit of volume.

5.5.1 Reynolds’ Lemma


Consider an arbitrary material volume of the continuous medium that at time t
occupies the volume in space Vt ≡ V . The amount of the generic property A in
the material volume Vt at time t is


rs
Q (t) = ρΨ dV . (5.30)

ee
Vt ≡V

s gin
The variation along time of the content of property A in the material volume

t d le En
Vt is given by the time derivative of Q (t), which using expression (5.17) of the
material derivative of a volume integral (with μ = ρΨ ) results in

r
ba
ge ro or
eS m
   

ci
d (ρΨ )
 d
f

ra
Q (t) = ρΨ dV = + ρΨ ∇ · v dV . (5.31)
C d P cs

b
dt dt

a
Vt ≡V μ
i
V
an an n
y ha

Considering the expression of the material derivative of a product of functions,


le
liv or ec

grouping terms and introducing the mass continuity equation (5.24) yields
   
M

.A

d dΨ dρ
ρΨ dV = ρ +Ψ + ρΨ ∇ · v dV =
dt dt dt
m

Vt ≡V V 
 
d


uu

dΨ dρ (5.32)
e

= ρ +Ψ + ρ∇ · v dV =⇒
X Th

er

dt dt
tin

V 
=0 (mass continuity eqn.)
on

.O
C

Reynolds’ Lemma
 
©

d
ρΨ dV = ρ

dV . (5.33)
dt dt
Vt ≡V V

5.5.2 Reynolds’ Theorem


Consider the arbitrary volume V , fixed in space, shown in Figure 5.8. The
amount of property A in this control volume is

Q (t) = ρΨ dV . (5.34)
V

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Balance Equation. Reynolds Transport Theorem 207

rs
ee
Figure 5.8: Reynolds Transport Theorem.

s gin
t d le En
The variation of the amount of property A in the material volume Vt , which in-

r
stantaneously coincides at time t with the control volume V (Vt ≡ V ), is given by

ba
ge ro or
eS m
expression (5.19) of the material derivative of a volume integral (with μ = ρΨ )

ci
f

ra
and by (5.11),
C d P cs
b
a
  
i
d ∂ (ρΨ )
an an n

ρΨ dV = dV + ∇ · (ρΨ v) dV . (5.35)
y ha

dt ∂t
Vt ≡V
le
V V
liv or ec

Introducing the Reynolds’ Lemma (5.33) and the Divergence Theorem10 in


M

.A

(5.35) results in
m

Reynolds’
uu

  
e

d Lemma dΨ ∂ (ρΨ )
X Th

ρΨ dV = ρ dV = dV + ∇ · (ρΨ v) dV =
er
tin

dt dt ∂t
Vt ≡V V V V
on

.O

Divergence  
Theorem ∂ (ρΨ )
= dV + ρΨ v · n dS ,
C

∂t
©

V ∂V
(5.36)
which can be rewritten as follows.

10The Divergence Theorem provides the following relation between a volume integral and a
surface integral of a tensor A.
 
∇ · A dV = n · A dS ∀V ,
V ∂V

where n is the outward unit normal vector in the boundary of the volume V .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
208 C HAPTER 5. BALANCE P RINCIPLES

Reynolds Transport Theorem


  
∂ ∂Ψ
ρΨ dV = ρ dV − ρΨ v · n dS
∂t ∂t (5.37)
V V ∂V
  
variation per unit of variation due to the variation due to the net
time of the content of change in the content of convective flux of A
property A in the property A of the parti- exiting through the
control volume V cles in the interior of V boundary ∂V

rs
The local form of the Reynolds Transport Theorem can be obtained by local-

ee
izing in (5.36),

s gin
  
dΨ ∂ (ρΨ )

t d le En
ρ dV = dV + ∇ · (ρΨ v) dV ∀ΔV ⊂ V =⇒
dt ∂t

r
V V V (5.38)

ba
ge ro or
eS m
dΨ ∂ (ρΨ )

ci
ρ = + ∇ · (ρΨ v) ∀x ∈ V =⇒
∂t f

ra
dt
C d P cs
b
a
i
an an n

Local form of the Reynolds Transport Theorem


y ha

(5.39)
∂ (ρΨ )
le

liv or ec

=ρ − ∇ · (ρΨ v) ∀x ∈ V
∂t dt
M

.A
m

d
uu

5.6 General Expression of the Balance Equations


e
X Th

er
tin

Consider a certain property A of a continuous medium and the amount of this


property per unit of mass, Ψ (x,t). In the most general case, it can be assumed
on

.O

that there exists an internal source that generates property A and that this prop-
C

erty can be transported both by motion of mass (convective transport) and by


©

non-convective transport. To this aim, the following terms are defined:


• A source term kA (x,t) (of the same tensor order than property A) that
characterizes the internal generation of the property,
internally generated amount of A
kA (x,t) = . (5.40)
unit of mass / unit of time
• A vector jA (x,t) of non-convective flux per unit of surface (a tensor
order higher than that of property A) that characterizes the flux of the
property due to non-convective mechanisms (see Remark 5.3).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
General Expression of the Balance Equations 209

rs
ee
s gin
t d le En
Figure 5.9: An arbitrary control volume used in the definition of the global form of the

r
general balance equation.

ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
Consider an arbitrary control volume V (see Figure 5.9). Then, the variation
i
per unit of time of property A in volume V will be due to
an an n
y ha

1) the generation of property A per unit of time due to the source term,
le
liv or ec

2) the (net incoming) convective flux of A through ∂V , and


M

.A

3) the (net incoming) non-convective flux of A through ∂V .


m

That is,
uu
e
X Th

er


tin

amount of A generated in V due to the internal sources


ρkA (x,t) dV = ,
on

unit of time
.O

V
C


amount of A exiting through ∂V per convective flux
©

ρΨ v · n dS = ,
unit of time
∂V

amount of A exiting through ∂V per non-convective flux
jA · n dS = ,
unit of time
∂V
(5.41)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
210 C HAPTER 5. BALANCE P RINCIPLES

and the expression of the balance of the amount of property A in the control
volume V results in

Global form of the general balance equation


   

ρΨ dV = ρkA dV − ρΨ v · n dS − jA · n dS
∂t
V V ∂V ∂V (5.42)
   
variation of the variation due variation due to variation due to
amount of A in V to internal the incoming the incoming

rs
per unit of time generation convective flux non-

ee
convective flux

s gin
Using the Divergence Theorem and (5.11), the global form of the general
balance equation (5.42) can be written as

t d le En
   

r

ba
ge ro or
ρΨ dV = ρkA dV − ∇ · (ρΨ v) dV − ∇ · jA dV =⇒

eS m
ci
∂t
f

ra
V V V V
   (5.43)
C d P cs

b
a

(ρΨ ) + ∇ · (ρΨ v) dV = (ρkA − ∇ · jA ) dV ∀ΔV ⊂ V
i
an an n

∂t
y ha

V V
le
liv or ec

and localizing in (5.43), the local spatial form of the general balance equation
M

.A
m

Local spatial form of the general balance equation


d
uu


e


(ρΨ ) + ∇ · (ρΨ v) = ρ = ρkA − ∇ · jA
X Th

er

∂t
tin

dt  
  (5.44)
dΨ variation of the variation variation
on

.O

ρ amount of due to due to non-


dt property (per internal convective
C

unit of volume generation transport


©

and of time) by a source

is obtained, where the local form of the Reynolds Transport Theorem (5.39) has
been taken into account.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
General Expression of the Balance Equations 211

Remark 5.5. Expression (5.42) and, especially, expression (5.44),



ρ = ρkA − ∇ · jA ,
dt
exhibit the negative contribution (−∇ · jA ) of the non-convective
flux to the variation in content of the property per unit of volume
and of time, ρ dΨ /dt. Only when all the flux is convective (by mass
transport) can this variation originate solely from the internal gener-

rs
ation of this property,

ee

ρ = ρkA .
dt

s gin
t d le En

r
ba
ge ro or
eS m
Example 5.2 – Particularize the local spatial form of the general balance

ci
f
equation for the case in which property A is associated with the mass.

ra
C d P cs
b
a
i
Solution
an an n
y ha

If property A is associated with the mass, A ≡ M, then:


le
liv or ec

• The content of A per unit of mass (mass / unit of mass) is Ψ = 1.


M

.A

• The source term that characterizes the internal generation of mass is


kM = 0 since, following the principle of conservation of mass, it is not
m

possible to generate mass.


uu
e

• The non-convective mass flux vector is jM = 0 because mass cannot be


X Th

er
tin

transported in a non-convective manner.


on

.O

Therefore, (5.44) results in the balance of mass generation,


C

dΨ ∂ρ
ρ = + ∇ · (ρv) = 0 ,
©

dt ∂t
which is one of the forms of the mass continuity equation (5.26).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
212 C HAPTER 5. BALANCE P RINCIPLES

5.7 Balance of Linear Momentum


Consider a discrete system composed of n particles
such that the particle i has a mass mi , an acceleration
ai and is subjected to a force fi (see Figure 5.10).
Newton’s second law states that the force acting
on a particle is equal to the mass of this particle
times its acceleration. Using the definition of accel-
eration as the material derivative of the velocity and
considering the principle of conservation of mass
(the variation of mass of a particle is null) yields11 , Figure 5.10

rs
ee
dvi d
fi = m i ai = m i = (mi vi ) (5.45)

s gin
dt dt
The linear momentum of the particle12 is defined as the product of its mass

t d le En
by its velocity (mi vi ). Then, (5.45) expresses that the force acting on the particle

r
is equal to the variation of the linear momentum of the particle.

ba
ge ro or
eS m
Applying now Newton’s second law to the discrete system formed by n par-

ci
ticles results in
f

ra
C d P cs
b
a
n n n
dvi d n
d P (t)
R (t) = ∑ fi = ∑ mi ai = ∑ mi ∑ mi vi
i
= = . (5.46)
an an n

dt dt dt
y ha

i=1 i=1 i=1 i=1



le
P = linear
liv or ec

momentum
M

.A

Note that, again, to obtain the last expression in (5.46), the principle of conser-
vation of mass (dmi /dt = 0) has been used. Equation (5.46) expresses that the
m

resultant R of all the forces acting on the discrete system of particles is equal
uu
e

to the variation per unit of time of the linear momentum P of the system. This
X Th

er
tin

postulate is denominated the principle of balance of linear momentum.


on

.O

Remark 5.6. If the system is in equilibrium, R = 0. Then,


C

d P (t) n
R (t) = 0 ∀t =⇒
dt
=0 =⇒ ∑ mi vi = P = const. ,
i=1

which is known as the conservation of linear momentum.

11 The Einstein notation introduced in (1.1) is not used here.


12In mechanics, the names translational momentum, kinetic momentum or simply momentum
are also used to refer to the linear momentum.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Balance of Linear Momentum 213

5.7.1 Global Form of the Balance of Linear Momentum


These concepts, corresponding to classical mechanics, can now be extended to
continuum mechanics by defining the linear momentum in a material volume Vt
of the continuous medium with mass M as
 
P (t) = v d M = ρ v dV . (5.47)

M ρ dV Vt

rs
Definition 5.5. Principle of balance of linear momentum. The resul-

ee
tant R (t) of all the forces acting on a material volume of the contin-

s gin
uous medium is equal to the variation per unit of time of its linear
momentum, 
d P (t)

t d le En
d
R (t) = = ρ v dV .
dt dt

r
ba
Vt

ge ro or
eS m
ci
f

ra
C d P cs
b
a
The resultant of all the forces acting on the continuous medium defined above
i
an an n

is also known to be (see Figure 5.11)


y ha

 
le
liv or ec

R (t) = ρb dV + t dS . (5.48)
M

.A

V ∂V
 
m

body surface
d

forces forces
uu
e
X Th

Applying the principle of balance of linear momentum on the resultant in (5.48)


er
tin

yields the integral form of the balance of linear momentum.


on

.O
C

Global form of the principle of balance of linear momentum


  
©

d (5.49)
ρb dV + t dS = ρv dV
dt
V ∂V Vt ≡V

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
214 C HAPTER 5. BALANCE P RINCIPLES

rs
ee
s gin
t d le En
Figure 5.11: Forces acting on a material volume of the continuous medium.

r
ba
ge ro or
eS m
ci
5.7.2 Local Form of the Balance of Linear Momentum f

ra
C d P cs
b
a
i
Using Reynolds’ Lemma (5.33) on (5.49) and introducing the Divergence The-
an an n

orem, results in
y ha

    ⎫
le
d dv
liv or ec

ρv dV = ρb dV + n · σ dS = ρ dV ⎪


dt  dt ⎪

M

.A

Vt ≡V V ∂V V ≡V ⎬
t t
 Divergence  =⇒ (5.50)
m

Theorem ⎪


d

n · σ dS = ∇ · σ dV ⎪

uu


e
X Th

∂V V
er
tin

 
dv
=⇒ (∇ · σ + ρb) dV + ρ ∀ΔV ⊂ V
on

.O

dV (5.51)
dt
C

V V
©

and, localizing in (5.51), yields the local spatial form of the balance of linear
momentum, also known as Cauchy’s equation13 .

Local spatial form of the principle of balance of linear momentum


(Cauchy’s equation)
(5.52)
dv
∇ · σ + ρb = ρ = ρa ∀x ∈ V, ∀t
dt

13 The Cauchy equation (already stated, but not deduced, in Chapter 4 ) is, thus, identified as
the local spatial form of the balance of linear momentum.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Balance of Angular Momentum 215

5.8 Balance of Angular Momentum


Consider a discrete system composed of n parti-
cles such that for an arbitrary particle i, its posi-
tion vector is ri , its mass is mi , a force fi acts on
it, and it has a velocity vi and an acceleration ai
(see Figure 5.12). The moment about the origin
of the force acting on this particle is Mi = ri × fi
and the moment about the origin of the linear
momentum14 of the particle is Li = ri × mi vi .
Considering Newton’s second law, the moment

rs
Figure 5.12
Mi is15

ee
dvi
Mi = ri × fi = ri × mi ai = ri × mi (5.53)

s gin
dt
Extending the previous result to the discrete system formed by n particles, the

t d le En
resultant moment about the origin MO of the forces acting on the system of
particles is obtained as16

r
ba
ge ro or
eS m

ci
n n n
dvi ⎪
MO (t) = ∑ ri × fi = ∑ ri × mi ai = ∑ ri × mif ⎪

ra

dt ⎪
C d P cs


b
a
i=1 i=1 i=1 ⎪


i
an an n

n n n
d dri dvi
∑ × = ∑ dt ×m + ∑ × =⇒
y ha

r m v v r m
dt ⎪
i i i i i i i
dt i=1 i=1  ⎪

le
i=1


liv or ec

vi ⎪

 ⎪

M

.A

(5.54)
=0
m

d n
dL (t)
uu

=⇒ MO (t) = ∑ ri × mi vi =
e
X Th

dt dt
er
tin

i=1

Angular
on

.O

momentum L
C

Equation (5.54) expresses that the resultant moment MO of all the forces act-
©

ing on the discrete system of particles is equal to the variation per unit of time
of the moment of linear momentum (or angular momentum), L , of the system.
This postulate is named principle of balance of angular momentum.

15 In mechanics, the moment of (linear) momentum is also named angular momentum or


rotational momentum.
15 The Einstein notation introduced in (1.1) is not used here.
16 The vector or cross product of a vector times itself is null (v × v = 0).
i i

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
216 C HAPTER 5. BALANCE P RINCIPLES

Remark 5.7. If the system is in equilibrium, MO = 0. Then,


dL (t) n
MO (t) = 0 ∀t =⇒ = 0 =⇒ ∑ ri × mi vi = L = const.,
dt i=1
which is known as the conservation of angular momentum.

5.8.1 Global Form of the Balance of Angular Momentum

rs
Result (5.54) can be extended to a continuous and infinite system of particles

ee
(the continuous medium, see Figure 5.13). In such case, the angular momentum

s gin
is defined as  
L = r × v d M = r × ρ v dV (5.55)

t d le En

M ρ dV V

r
ba
ge ro or
eS m
and the continuous version of the postulate of balance of angular momentum is

ci
f

ra
obtained as follows.
C d P cs
b
a
i
an an n
y ha

Definition 5.6. Principle of balance of moment of (linear) momen-


le
tum or angular momentum. The resultant moment, about a certain
liv or ec

point O in space, of all the actions on a continuous medium is equal


M

.A

to the variation per unit of time of the moment of linear momentum


about said point.
m

dL (t) d 
d

MO (t) = = r × ρ v dV
uu
e

dt dt Vt ≡V
X Th

er
tin
on

.O
C

Figure 5.13: Moments acting on a material volume of the continuous medium.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Balance of Angular Momentum 217

The resultant moment of the forces acting on the continuous medium (mo-
ment of the body forces and moment of the surface forces) is (see Figure 5.13)
 
MO (t) = r × ρ b dV + r × t dS , (5.56)
V ∂V

then, the global form of the principle of balance of the angular momentum re-
sults in:

Global form of the principle of balance of angular momentum

rs
  

ee
d (5.57)
r × ρ v dV = r × ρ b dV + r × t dS
dt

s gin
Vt ≡V V ∂V

t d le En

r
ba
ge ro or
5.8.2 Local Spatial Form of the Balance of Angular Momentum

eS m
ci
f

ra
The procedure followed to obtain the local spatial form of the balance equation
C d P cs
b
a
is detailed below.
i
an an n

Introducing Reynolds’ Lemma in (5.57),


y ha

  
le
liv or ec

d d d
r × ρv dV = ρ (r × v) dV = ρ (r × v) dV =
M

.A

dt dt dt
Vt ≡V Vt ≡V V
  dr     
m

dv dv (5.58)
= ρ ×v dV + ρ r × dV = r × ρ dV ,
d
uu

dt dt dt
e

V  V V
X Th

er

v
tin


=0
on

.O

and expanding the last term in (5.57),


C

  
r × t dS = r × n · σ dS = [r] × [n · σ ]T dS =

∂V n·σ ∂V ∂V
(5.59)
 Divergence
 
Theorem
= (r × σ ) · n dS
T
= r × σ T · ∇ dV ,
∂V V

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
218 C HAPTER 5. BALANCE P RINCIPLES

  
where the component r × σ T · ∇ i is computed as

  symb
  ∂ ∂  
r×σT ·∇ = ei jk x j σrk = ei jk x j σrk =
i  ∂ xr ∂ xr
σkr
T
(5.60)
∂xj ∂ σrk
= ei jk σrk + ei jk x j = ei jk σ jk + [r × ∇ · σ ]i i ∈ {1, 2, 3} .
∂ xr ∂ xr
  
[r × ∇ · σ ]i mi
δ jr

rs
ee
Introducing now (5.60) in (5.59) produces

s gin
  
r × t dS = m dV + (r × ∇ · σ ) dV

t d le En
∂V V V (5.61)

r
ba
ge ro or
eS m
mi = ei jk σ jk i, j, k ∈ {1, 2, 3}

ci
f

ra
C d P cs
b
a
i
and, finally, replacing (5.58) and (5.61) in (5.57) yields
an an n

   
y ha

dv
r×ρ dV = r × ρb dV + m dV + (r × ∇ · σ ) dV .
le
(5.62)
liv or ec

dt
V V V V
M

.A

Reorganizing the terms in (5.62) and taking into account Cauchy’s equation (5.52)
m

(local spatial form of the balance of linear momentum) results in


d
uu

 
e

 
X Th

dv
er

r × ∇ · σ + ρb − ρ dV + m dV = 0
tin

dt
V  V
on

.O

=0  (5.63)
C

=⇒ m dV = 0 ∀ΔV ⊂ V .
©

Then, localizing in (5.63) and considering the value of m in (5.61), yields


!
m = 0 ∀x ∈ V
=⇒ ei jk σ jk = 0 i, j, k ∈ {1, 2, 3} (5.64)
mi = ei jk σ jk = 0 i ∈ {1, 2, 3}

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Power 219

and particularizing (5.64) for the three possible values of index i:



i = 1 : e1 jk σ jk = e123 σ23 + e132 σ32 = σ23 − σ32 = 0 ⇒ σ23 = σ32 ⎪


  ⎪

=1 =−1 ⎪



i = 2 : e2 jk σ jk = e231 σ31 + e213 σ13 = σ31 − σ13 = 0 ⇒ σ31 = σ13
  =⇒
=1 =−1 ⎪



i = 3 : e3 jk σ jk = e312 σ12 + e321 σ21 = σ12 − σ21 = 0 ⇒ σ12 = σ21 ⎪


  ⎪

=1 =−1
=⇒ σ = σT ,

rs
(5.65)

ee
which results in the local spatial form of the balance of angular momentum

s gin
translating into the symmetry of the Cauchy stress tensor17 .

t d le En
Local spatial form of the

r
principle of balance of angular momentum

ba
ge ro or
(5.66)

eS m
ci
σ = σT
f

ra
C d P cs
b
a
i
an an n
y ha

le
5.9 Power
liv or ec
M

.A
m

Definition 5.7. In classical mechanics as well as in continuum me-


uu

chanics, power is defined as a concept, previous to that of energy,


e
X Th

er

that can be quantified as the ability to perform work per unit of time.
tin

Then, for a system (or continuous medium) the power W (t) entering
on

.O

the system is defined as


C

work performed by the system


W (t) = .
©

unit of time

In some cases, but not in all, the power W (t) is an exact differential of a function
E (t) that, in said cases, receives the name of energy,
d E (t)
W (t) = . (5.67)
dt

17 The symmetry of the Cauchy stress tensor (already stated, but not deduced, in Chapter 4 )
is, thus, identified as the local spatial form of the balance of angular momentum.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
220 C HAPTER 5. BALANCE P RINCIPLES

Here, it is assumed that there exist two procedures by which the continuous
medium absorbs power from the exterior and performs work per unit of time
with this power
− Mechanical power, by means of the work performed by the mechanical
actions (body and surface forces) acting on the medium.
− Thermal power, by means of the heat entering the medium.

5.9.1 Mechanical Power. Balance of Mechanical Energy

rs
ee
s gin
Definition 5.8. The mechanical power entering the continuous
medium, Pe , is the work per unit of time performed by all the (body

t d le En
and surface) forces acting on the medium.

r
ba
ge ro or
eS m
ci
f
Consider the continuous medium shown in Figure 5.14 is subjected to the ac-

ra
C d P cs
tion of body forces, characterized by the vector of body forces b (x,t), and of
b
a
surface forces, characterized by the traction vector t (x,t). The expression of the
i
an an n

mechanical power entering the system Pe is


y ha

   
le
liv or ec

Pe = ρ b · v dV + σ · v) dS .
t · v dS = ρ b · v dV + n · (σ (5.68)

M

.A

V ∂V n·σ V ∂V
m

d
uu
e
X Th

er
tin
on

.O
C

dr
ρb · dV = ρb · vdV
©

dt

dr
t· dS = t · vdS
dt

Figure 5.14: Continuous medium subjected to body and surface forces.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Power 221

Applying the Divergence Theorem in the last term of (5.68) yields


⎧ 

⎪ σ · v) dS = ∇ · (σ
n · (σ σ · v) dV



⎨∂V V
∂ ∂ σi j ∂vj

⎪ σ · v) =
∇ · (σ (σi j v j ) = v j + σi j = (∇ · σ ) · v + σ : l

⎪ ∂ xi ∂ xi ∂ xi

⎩   
[∇ · σ ] j σ ji [ll ]
ji
(5.69)
and, taking into account the identity l = v ⊗ ∇ = d + w (see Chapter 2)18 ,

rs
σ : l = σ : d+σ : w = σ : d . (5.70)
 

ee
d+w =0

s gin
Replacing (5.70) in (5.69) yields

t d le En
  
σ · v) dS =
n · (σ (∇ · σ ) · v dV + σ : d dV . (5.71)

r
ba
ge ro or
eS m
∂V V V

ci
f

ra
C d P cs
Introducing (5.71) in (5.68), the mechanical power entering the continuous
b
a
results in19 
i
medium
an an n

   
y ha

Pe = ρ b · v dV + t · v dS = ρ b · v dV + (∇ · σ ) · v dV + σ : d dV =
le
liv or ec

 V ∂V  V  V  V
dv
M

.A

= (∇ · σ + ρ b) · v dV + σ : d dV = ρ · v dV + σ : d dV =
dt
   V  
m

 V  V V
d

d 1 d 1 2
ρ v · v dV + σ : d dV = ρ v dV + σ : d dV .
uu
e

dt 2 dt 2
X Th

er

V V V V
tin

(5.72)
And applying Reynolds’ Lemma (5.33) in (5.72), the mechanical power entering
on

.O

the system results in


C

Balance of mechanical energy


©

   
d 1 2
Pe = ρ b · v dV + t · v dS = ρv dV + σ : d dV
dt 2 (5.73)
 V ∂V Vt ≡V V
mechanical  
power entering K = kinetic stress
the medium energy power

18 The tensor σ is symmetric and the tensor w is antisymmetric. Consequently, their product
σ : w = 0.
is null,σ  
19 The expression d 1
v·v =
1 dv 1 dv dv
·v+ v· = · v is used here, in addition to the
dt 2 2 dt 2 dt dt
notation v · v = |v|2 = v2 .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
222 C HAPTER 5. BALANCE P RINCIPLES

Equation (5.73) constitutes the continuum mechanics generalization of the


balance of mechanical energy in classical mechanics.

Definition 5.9. The balance of mechanical energy states that the me-
chanical energy entering the continuous medium,
 
Pe = ρ b · v dV + t · v dS
V ∂V
is invested in:
a) modifying the kinetic energy of the particles in the continuous

rs
medium,  
dK d

ee
not 1 2 1 2
kinetic energy = K = ρ v dV =⇒ = ρ v dV .
2 dt dt 2

s gin
V V
b) creating stress power,


t d le En
de f
stress power = σ : d dV .

r
ba
ge ro or
V

eS m
ci
f

ra
C d P cs
b
a
i
Remark 5.8. Considering (5.73), the stress power can be defined as
an an n
y ha

the part of the mechanical power entering the system that is not used
in modifying the kinetic energy. It can be interpreted as the work per
le
liv or ec

unit of time (power) performed by the stresses during the deforma-


M

.A

tion process of the medium.


In a rigid body there is no deformation nor strain rate (d = 0). There-
m

fore, the stresses do not perform mechanical work and the stress
d
uu

power is null. In this case, all the mechanical power entering the
e
X Th

system is invested in modifying the kinetic energy of the system and


er
tin

the balance of mechanical energy of a rigid body is recovered.


on

.O
C

5.9.2 Thermal Power

Definition 5.10. The thermal power entering the continuous


medium, Qe , is the amount of heat per unit of time entering the
medium.

The heat entering the medium can be produced by two main causes:
a) Heat entering the medium due to the (non-convective) heat flux across the
boundary corresponding to the material volume. Note that, since the vol-

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Power 223

ume is a material volume, the heat flux due to mass transport (convective) is
null and, thus, all the heat flux entering the medium will be non-convective.
b) The existence of heat sources inside the continuous medium.
• Non-convective heat flux
Consider the spatial description of the vector of non-convective heat flux
per unit of surface, q (x,t). Then, the net non-convective heat flux across
the boundary of the material volume is (see Figure 5.15)

amount of heat exiting the medium

rs
q · n dS =
unit of time

ee
∂V  (5.74)

s gin
amount of heat entering the medium
− q · n dS =
unit of time

t d le En
∂V

r
ba
ge ro or
eS m
ci
f
Remark 5.9. A typical example of non-convective flux is heat trans-

ra
C d P cs
fer by conduction phenomena. Heat conduction is governed by
b
a
i
Fourier’s Law, which provides the vector of heat flux by (non-
an an n

convective) conduction q (x,t) in terms of the temperature θ (x,t),


y ha


le
liv or ec

Fourier’s Law of q (x,t) = −K∇θ (x,t) ,


heat conduction
M

.A

where K is the thermal conductivity, a material property.


m

d
uu
e
X Th

er
tin
on

.O
C

Figure 5.15: Non-convective heat flux.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
224 C HAPTER 5. BALANCE P RINCIPLES

rs
ee
Figure 5.16: Internal heat sources.

s gin
t d le En
• Internal heat sources

r
ba
ge ro or
eS m
Heat can be generated (or absorbed) in the interior of the continuous

ci
f
medium due to certain phenomena (chemical reactions, etc.). Consider a

ra
C d P cs
scalar function r (x,t) that describes in spatial form the heat generated by
b
a
i
the internal sources per unit of mass and unit of time (see Figure 5.16).
an an n
y ha

Then, the heat entering the system, per unit of time, due to the existence of
internal heat sources is
le
liv or ec


heat generated by the internal sources
M

.A

ρr dV = . (5.75)
unit of time
m

V
d
uu

Consequently, the total heat entering the continuous medium per unit of
e
X Th

time (or thermal power Qe ) can be expressed as the sum of the contributions
er
tin

of the conduction flux (5.74) and the internal sources (5.75),


on

.O

Heat power entering  


C

Qe = ρr dV − q · n dS . (5.76)
the medium
©

V ∂V

Then, considering (5.73) and (5.76), the total power entering the continu-
ous medium can be written as follows.

Total power entering the system


   
d 1 2 (5.77)
Pe + Qe = ρv dV + σ : d dV + ρr dV − q · n dS
dt 2
Vt ≡V V V ∂V

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Energy Balance 225

5.10 Energy Balance


5.10.1 Thermodynamic Concepts
• Thermodynamic system: a certain amount of continuous matter always
formed by the same particles (in the case studied here, a material volume).
• Thermodynamic variables: a set of macroscopic variables that characterize
the system and intervene in all the physical processes to be studied. They
are designated by μi (x,t) i ∈ {1, 2, ... , n}.
• State, independent or free variables: a subset of the group of thermody-
namic variables in terms of which all the other variables can be expressed.

rs
• Thermodynamic state: a thermodynamic state is defined when a certain

ee
value is assigned to the state variables and, therefore, to all the thermo-

s gin
dynamic variables. In a hyperspace (thermodynamic space) defined by the
thermodynamic variables μi i ∈ {1, 2, ... , n} (see Figure 5.17), a thermo-

t d le En
dynamic state is represented by a point.
• Thermodynamic process: the energetic development of a thermodynamic

r
ba
ge ro or
eS m
system that undergoes successive thermodynamic states, changing from an

ci
f
initial state at time tA to a final state at time tB (it is a path or continuous

ra
C d P cs
segment in the thermodynamic space), see Figure 5.18.
b
a
• Closed cycle: A thermodynamic process in which the final thermodynamic
i
an an n

state coincides with the initial thermodynamic state (all the thermodynamic
y ha

variables recover their initial value), see Figure 5.19.


le
liv or ec

• State function: any scalar, vector or tensor function φ (μ1 , ... , μn ) of the
M

.A

thermodynamic variables that can be written univocally in terms of these


variables.
m

Consider a thermodynamic space with thermodynamic variables μi (x,t)


d
uu

i ∈ {1, 2, ... , n} and a function φ (μ1 , ... , μn ) of said variables implicitly defined
e
X Th

er

in terms of a differential form20


tin

δ φ = f1 (μ1 , ... , μn ) dμ1 + . . . + fn (μ1 , ... , μn ) dμn .


on

.O

(5.78)
C

Figure 5.17: Thermodynamic process.

20 In continuum mechanics thermodynamics it is common to mathematically describe a func-


tion φ (μ1 , ... , μn ) of the thermodynamic variables in terms of a differential form δ φ .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
226 C HAPTER 5. BALANCE P RINCIPLES

Figure 5.18: Thermodynamic space.

rs
Figure 5.19: Closed cycle.

ee
s gin
Consider also a given thermodynamic process A → B in the space of the
thermodynamic variables. Equation (5.78) provides the value of the function

t d le En
not not
φ (μ1B , ... , μnB ) = φB when its value φ (μ1A , ... , μnA ) = φA and the corresponding

r
path (thermodynamic process) A → B are known by means of

ba
ge ro or
eS m
ci
f
B

ra
C d P cs
φB = φA + δφ .
b
a
(5.79)
i
an an n

A
y ha

However, (5.79) does not guarantee that the result φB is independent of the path
le
liv or ec

(thermodynamic process) followed. In mathematical terms, it does not guarantee


that the function φ : Rn → R defined by (5.79) is univocal (see Figure 5.20) and,
M

.A

thus, that there exists a single image φ (μ1 , ... , μn ) corresponding to each point
m

in the thermodynamic space.


d
uu
e
X Th

er
tin
on

.O
C

Figure 5.20: Non-univocal function of the thermodynamic variables μ1 and μ2 .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Energy Balance 227

Remark 5.10. For a function φ (μ1 , ... , μn ), implicitly described in


terms of a differential form δ φ , to be a state function (that is, for it
to be univocal), said differential form must be an exact differential
δ φ = dφ . In other words, the differential form δ φ must be inte-
grable.
The necessary and sufficient condition for a differential form such
as (5.78) to be an exact differential is the equality of mixed partial
derivatives,

rs

δ φ = f1 (μ1 , ... , μn ) dμ1 + . . . + fn (μ1 , ... , μn ) dμn ⎪ ⎬

ee
∂ fi (μ1 , ... , μn ) ∂ f j (μ1 , ... , μn ) ⇔ δ φ = dφ .

s gin
= ∀i, j ∈ {1, ... , n} ⎪


∂ μj ∂ μi

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
If the differential form (5.78) is an exact differential, (5.79) results in
b
a
i
an an n

B 
y ha

B
φ B = φA + dφ = φA + Δ φ (5.80)
le
liv or ec

A
A
M

.A

and the value φB is independent of the integration path. Then, function φ is said
m

to be a state function that depends only on the values of the state variables and
d
uu

not on the thermodynamic process.


e
X Th

er
tin
on

.O

Remark 5.11. If φ is a state function, then δ φ is an exact differential


C

and the integral along the complete closed cycle of the differential
δ φ is null,
©

A "  A
δ φ = dφ = Δ φ = 0 .
A A
=0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
228 C HAPTER 5. BALANCE P RINCIPLES

Example 5.3 – Determine whether the function φ (μ1 , μ2 ) defined in terms of


an exact differential δ φ = 4μ2 dμ1 + μ1 dμ2 can be a state function or not.

Solution
Following (5.78),
∂ f1
f1 ≡ 4μ2 =4 ∂ f1 ∂ f2
=⇒ ∂ μ2 =⇒ =
f 2 ≡ μ1 ∂ f2 ∂ μ 2 ∂ μ1
=1
∂ μ1

rs
Then, δ φ is not an exact differential (see Remark 5.10) and φ is not a state

ee
function.

s gin
t d le En
5.10.2 First Law of Thermodynamics

r
ba
ge ro or
Experience shows that the mechanical power (5.73) is not an exact differential

eS m
ci
and, therefore, the mechanical work performed by the system in a closed cycle
f

ra
C d P cs
is not null. The same happens with the thermal power (5.76).
b
a
"
i
an an n

δ φ1 = Pe dt =⇒ Pe dt = 0
y ha

" (5.81)
le
δ φ2 = Qe dt =⇒ Qe dt = 0
liv or ec
M

.A

However, there exists experimental evidence that proves that the sum of the me-
m

chanical and thermal powers, that is, the total power entering the system (5.77)
d

(see Figure 5.21), is, in effect, an exact differential and, thus, a state function E
uu
e

that corresponds to the concept of energy can be defined in terms of it,


X Th

er
tin

t
on

.O

Pe dt + Qe dt = d E ⇒ E (t) = (Pe + Qe ) dt + const. (5.82)


C

t0
©

Figure 5.21: Total power entering the system.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Energy Balance 229

The first law of thermodynamics postulates the following:


1) There exists a state function E , named total energy of the system, such that
its variation per unit of time is equal to the sum of the mechanical and
thermal powers entering the system.

dE
= Pe + Qe
dt
dE = Pe dt + Qe dt (5.83)
  

rs
Variation of Mechanical Thermal
total energy work work

ee
s gin
2) There exists another state function U , named internal energy of the system,
such that

t d le En
a) It is an extensive property21 . Then, a specific internal energy u (x,t)
(or internal energy per unit of mass) can be defined as

r
ba
ge ro or


eS m
ci
U=f ρu dV . (5.84)

ra
C d P cs
b
a
V
i
an an n

b) The variation of the total energy of the system E is equal to the sum of
y ha

the variation of the internal energy U and the variation of the kinetic
le
liv or ec

energy K.
M

.A

dE = dK + dU
  (5.85)
m

Exact Exact
uu

differential differential
e
X Th

er
tin

Remark 5.12. Note that, since the total energy E and the internal en-
on

.O

ergy U of the system have been postulated to be state functions,


C

d E and d U in (5.85) are exact differentials. Consequently, the term


©

d K = d E − d U in said equation is also an exact differential (because


the difference between exact differentials is also an exact differen-
tial) and, thus, is a state function. Then, it is confirmed that (5.85)
indirectly postulates the character of state function and, therefore,
the energetic character of K.

21 A certain property is extensive when the complete content of the property is the sum of the
content of the property in each of its parts. An extensive property allows defining the content
of this property per unit of mass (specific value of the property) or per unit of volume (density
of the property).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
230 C HAPTER 5. BALANCE P RINCIPLES

From (5.83) and considering (5.77),


    ⎫
dE d 1 2 ⎪
= Pe + Qe = ρv dV + σ : d dV + ρr dV − q · n dS ⎪


dt dt 2 ⎬
Vt ≡V ∂V
 V V

1 2 ⎪

K= ρv dV ⎪

2 ⎭
V

   
dE dK dU d 1 2
= + = ρv dV + σ : d dV + ρr dV − q · n dS
dt dt dt dt 2

rs
V V V ∂V
 

ee
dK dU

s gin
dt dt
(5.86)

t d le En

r
ba
ge ro or
eS m
ci
Global form of the internal energy balance f

ra
C d P cs
   
b
a
dU d (5.87)
i
= ρu dV = σ : d dV + ρr dV − q · n dS
an an n

dt dt
y ha

Vt ≡V V V ∂V
le
liv or ec
M

.A

Remark 5.13. From (5.87) it follows that any variation per unit of
m

time of the internal energy d U /dt is produced by


d
uu
e


X Th

er

− a generation of stress power, σ : d dV , and


tin

V
on

.O

− a variation per
 unit of time of the content of heat in the medium,
C

ρr dV − q · n dS.
©

V ∂V

Applying Reynolds’ Lemma (5.33) and the Divergence Theorem on (5.87)


yields
    
d du
ρu dV = ρ dV = σ : d dV + ρr dV − ∇ · q dV ∀ΔV ⊂ V .
dt dt
Vt ≡V V V V V
(5.88)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Reversible and Irreversible Processes 231

Finally, localizing in (5.88) results in the local spatial form of the internal energy
balance.

Local spatial form of the internal energy balance


(energy equation)
(5.89)
du
ρ = σ : d + (ρr − ∇ · q) ∀x ∈ V, ∀t
dt

rs
5.11 Reversible and Irreversible Processes

ee
s gin
The first law of thermodynamics leads to a balance equation that must be ful-
filled for all the physical processes that take place in reality,

t d le En
dE dU dK
Pe + Qe = = + .

r
(5.90)

ba
ge ro or
dt dt dt

eS m
ci
f

ra
In particular, if an isolated system22 is considered, the time variation of the total
C d P cs
b
energy of the system will be null (d E /dt = 0 ⇒ the total energy is conserved).

a
i
an an n

Therefore, the energy balance equation (5.90), established by the first law of
thermodynamics, imposes that any variation of internal energy d U /dt must be
y ha

compensated with a variation of kinetic energy d K/dt of equal value but of


le
liv or ec

opposite sign, and vice-versa (see Figure 5.22).


M

.A

What the first law of thermodynamics does not establish is whether this (ki-
netic and internal) energy exchange in an isolated system can take place equally
m

in both directions or not (d U /dt = −d K/dt > 0 or d U /dt = −d K/dt < 0). That
d
uu

is, it does not establish any restriction that indicates if an imaginary and arbitrary
e
X Th

er
tin
on

.O
C

Figure 5.22: Isolated thermodynamic system.

22 An isolated thermodynamic system is a system that cannot exchange energy with its
exterior. In a strict sense, the only perfectly isolated system is the universe, although one can
think of quasi-isolated or imperfectly isolated smaller systems.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
232 C HAPTER 5. BALANCE P RINCIPLES

process that implies an energy exchange in a certain direction is physically pos-


sible or not. It only establishes the fulfillment of the energy balance (5.90) in the
event that the process takes place.
However, experience shows that certain pro-
cesses that could be imagined theoretically
never take place in reality. Suppose, for exam-
ple, the isolated system in Figure 5.23 consist-
ing of
− a rigid (non-deformable) wheel that spins
with angular velocity ω, and

rs
− a brake that can be applied on the wheel at

ee
a certain instant of time.

s gin
Figure 5.23

t d le En
Consider now the following two processes:

r
1) At a certain instant of time the brake acts, the rotation speed of the wheel ω

ba
ge ro or
eS m
decreases and, thus, so does its kinetic energy (d K < 0). On the other hand,

ci
f
due to the friction between the brake and the wheel, heat is generated and

ra
C d P cs
there is an increase of the internal energy (d U > 0). Experience shows
b
a
i
that this process, in which the internal energy increases at the expense of
an an n

decreasing the kinetic energy23 , can take place in reality and, therefore, is
y ha

a physically feasible process.


le
liv or ec

2) Maintaining the brake disabled, at a certain instant of time the wheel spon-
M

.A

taneously increases its rotation speed ω and, thus, its kinetic energy in-
creases (d K > 0). According to the first law of thermodynamics, the in-
m

ternal energy of the system will decrease (d U < 0). However, experience
d
uu
e

shows that this (spontaneous) increase of speed never takes place, and nei-
X Th

er
tin

ther does the decrease in the amount of heat of the system (which would be
reflected in a decrease in temperature).
on

.O

The conclusion to this observation is that the second process considered in


C

the example is not a feasible physical process. More generally, only thermo-
©

dynamic processes that tend to increase the internal energy and decrease the
kinetic energy, and not the other way round, are feasible for the system under
consideration.
It is concluded, then, that the first law of thermodynamics is only applicable
when a particular physical process is feasible, and the need to determine when a
particular physical process is feasible, or if a physical process is feasible in one
direction, in both or in none, is noted. The answer to this problem is provided
by the second law of thermodynamics.
23 The wheel, being a non-deformable medium, has null stress power (see Remark 5.8) and
all the variation of internal energy of the system derives from a variation of its heat content
(see Remark 5.13).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Second Law of Thermodynamics. Entropy 233

Figure 5.24: Reversible (left) and irreversible (right) processes.

rs
ee
The previous considerations lead to the classification, from a thermodynamic

s gin
point of view, of the possible physical processes in feasible and non-feasible pro-
cesses and, in addition, suggest classifying the feasible processes into reversible

t d le En
and irreversible processes.

r
ba
ge ro or
eS m
ci
f
Definition 5.11. A thermodynamic process A → B is a reversible

ra
C d P cs
b
a
process when it is possible to return from the final thermodynamic
i
an an n

state B to the initial thermodynamic state A along the same path (see
y ha

Figure 5.24).
A thermodynamic process A → B is an irreversible process when
le
liv or ec

it is not possible to return from the final thermodynamic state B to


M

.A

the initial thermodynamic state A, along the same path (even if a


different path can be followed, see Figure 5.24).
m

d
uu
e
X Th

er
tin

In general, within a same thermodynamic process there will exist reversible


and irreversible sections.
on

.O
C

5.12 Second Law of Thermodynamics. Entropy


©

5.12.1 Second Law of Thermodynamics. Global form


The second law of thermodynamic postulates the following:
1) There exists a state function named absolute temperature θ (x,t) that is
intensive24 and strictly positive (θ > 0).
24 A certain property is intensive when the complete content of the property is not the sum
of the content of the property in each of its parts. Contrary to what happens with extensive
properties, in this case the content of the property cannot be defined per unit of mass (spe-
cific value of the property) or per unit of volume (density of the property). Temperature is a
paradigmatic example of intensive property.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
234 C HAPTER 5. BALANCE P RINCIPLES

2) There exists a state function named entropy S with the following character-
istics:
a) It is an extensive variable. This implies that there exists a specific
entropy (entropy per unit of mass) s such that

entropy
s= =⇒ S= ρs dV . (5.91)
unit of mass
V

b) The inequality

rs
Integral form of the second law of thermodynamics

ee
  
dS d r q (5.92)
= ρs dV ≥ ρ dV − · n dS

s gin
dt dt θ θ
Vt ≡V V ∂V

t d le En

r
is satisfied, where:

ba
ge ro or
eS m
− The sign = corresponds to reversible processes.

ci
f
− The sign > corresponds to irreversible processes.

ra
C d P cs
b
a
− The sign < cannot occur and indicates that the corresponding pro-
i
an an n

cess is not feasible.


y ha

le
5.12.2 Physical Interpretation of the Second Law of Thermodynamics
liv or ec
M

.A

As discussed Section 5.9.2, the magnitude heat in the system is characterized by


a) A source term (or generation of heat per unit of mass and unit of time)
m

r (x,t), defined in the interior of the material volume.


uu
e
X Th

b) The non-convective flux (heat flux by conduction) across the boundary of


er
tin

the material surface, defined in terms of a non-convective flux vector per


unit of surface q (x,t).
on

.O

These terms allow computing the amount of heat per unit of time entering a
C

material volume Vt , which at a certain instant of time occupies the spatial volume
©

Vt ≡ V with outward unit normal vector n, as


 
Qe = ρr dV − q · n dS . (5.93)
V ∂V

Consider now a new magnitude defined as heat per unit of absolute temper-
ature in the system. If θ (x,t) is the absolute temperature, the amount of said
magnitude will be characterized by
a) A source term r/θ corresponding to the generation of heat per unit of ab-
solute temperature, per unit of mass and unit of time.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Second Law of Thermodynamics. Entropy 235

b) A non-convective flux vector q/θ of the heat per unit of absolute temper-
ature.

Magnitude Source term Non-convective


flux vector
heat
r q
unit of time
heat/unit of absolute temperature r q

rs
unit of time θ θ

ee
s gin
Similarly to (5.93), the new source term r/θ and non-convective flux vec-
tor q/θ allow computing the amount of heat per unit of absolute temperature

t d le En
entering the material volume per unit of time as

r
 

ba
ge ro or
(heat/unit of temperature) entering V r q

eS m
= ρ dV − · n dS .

ci
(5.94)
unit of time f θ θ

ra
C d P cs
V ∂V

b
a
i
an an n

Observing now (5.94), the second term in this expression is identified as the
y ha

magnitude defined in (5.92). This circumstance allows interpreting the second


le
liv or ec

law of thermodynamics establishing that the generation of entropy per unit of


time in a continuous medium is always larger than or equal to the amount of
M

.A

heat per unit of temperature entering the system per unit of time.
m

d
uu
e
X Th

Global form of the second law of thermodynamics


er
tin

 
dS r q
≥ ρ dV − · n dS
on

.O

dt θ θ (5.95)
∂V
C

V

©

amount of the property


“heat / unit of absolute temperature”
entering the domain V per unit of time

Consider now the decomposition of the total entropy of the system S into two
distinct components:
• S(i) : entropy generated (produced) internally by the continuous medium. Its
generation rate is dS(i) /dt.
• S(e) : entropy generated by the interaction of the continuous medium with
its exterior. Its variation rate is dS(e) /dt.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
236 C HAPTER 5. BALANCE P RINCIPLES

Then, the following is naturally satisfied.

dS dS(e) dS(i)
= + (5.96)
dt dt dt

Now, if one establishes that the variation rate of the entropy generated by
the interaction with the exterior coincides with the magnitude heat per unit of
absolute temperature in (5.93),
 
dS(e) r q

rs
= ρ dV − · n dS (5.97)
dt θ θ

ee
V ∂V

s gin
and, taking into account (5.95) to (5.97), the variation per unit of time of the

t d le En
internally generated entropy results in
⎛ ⎞

r
ba
 

ge ro or
dS(i) dS dS(e) dS ⎝

eS m
r q
= − = − ρ dV − · n dS⎠ ≥ 0 .

ci
(5.98)
θ θ f

ra
dt dt dt dt
C d P cs
V ∂V
b
a
i
an an n
y ha

le
Remark 5.14. According to (5.98), the internally generated en-
liv or ec

tropy S(i) of the system (continuous medium) never decreases


M

.A

(dS(i) /dt ≥ 0). In a perfectly isolated system (strictly speaking, only


the universe is a perfectly isolated system) there is no interaction
m

with the exterior and the variation of entropy due to interaction with
uu
e

the exterior is null, (dS(e) /dt = 0). In this case, the second law of
X Th

er
tin

thermodynamics establishes that


dS(i) dS
on

.O

= ≥0
dt dt
C

or, in other words, the total entropy of a perfectly isolated system


©

never decreases. This is the starting point of some alternative for-


mulations of the second law of thermodynamics.

5.12.3 Reformulation of the Second Law of Thermodynamics


In view of the considerations in Section 5.12.2, the second law of thermodynam-
ics can be reformulated as follows:
1) There exists a state function named absolute temperature such that it is
always strictly positive,
θ (x,t) > 0 . (5.99)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Second Law of Thermodynamics. Entropy 237

2) There exists a state function named entropy that is an extensive variable


and, thus, can be defined in terms of a specific entropy (or entropy per unit
of mass) s (x,t) as 
S (t) = ρs dV . (5.100)
V

3) Entropy can be generated internally, S(i) , or produced by interaction with


the exterior, S(e) . Both components of the entropy are extensive variables
and their content in a material volume V can be defined in terms of their
respective specific values s(i) and s(e) ,

rs
 

ee
S(i) = ρs(i) dV and S(e) = ρs(e) dV (5.101)

s gin
V V

t d le En
dS dS(i) dS(e)
S = S(i) + S(e) = =⇒
+ (5.102)

r
dt dt dt

ba
ge ro or
eS m
and introducing Reynolds’ Lemma (5.33) in (5.102) yields

ci
f

ra
C d P cs
 
dS(i) ds(i)
b
a
d
= ρs(i) dV = ρ dV ,
i
an an n

dt dt dt
y ha

Vt ≡V V
  (5.103)
dS(e) ds(e)
le
d
liv or ec

(e)
= ρs dV = ρ dV .
dt dt dt
M

.A

Vt ≡V V
m

4) The variation of external entropy (generated by the interaction with the


d
uu

exterior) is associated with the variation of the magnitude heat per unit of
e
X Th

absolute temperature, and is defined as


er
tin

 
dS(e) r q
on

.O

= ρ dV − · n dS . (5.104)
dt θ θ
C

V ∂V
©

5) The internally generated entropy never diminishes. Based on the variation


of its content during the thermodynamic process, the following situations
are defined:

dS (i) ⎨ = 0 reversible process
≥ 0 → > 0 irreversible process (5.105)
dt ⎩
< 0 non-feasible process

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
238 C HAPTER 5. BALANCE P RINCIPLES

5.12.4 Local Form of the Second Law of Thermodynamics.


Clausius-Planck Equation
Using (5.101) to (5.104), expression (5.105) is rewritten as

dS(i) dS dS(e)
= − ≥0
dt dt dt
 
 
 (5.106)
d d r q
ρs(i) dV = ρs dV − ρ dV − · n dS ≥ 0
dt dt θ θ
Vt ≡V Vt ≡V ∂V

rs
V

ee
Applying Reynolds’ Lemma (5.33) (on the first and second integral of the left-

s gin
hand term in (5.106)) and the Divergence Theorem (on the last term) yields
 
 

ds(i) q

t d le En
ds r
ρ dV = ρ dV − ρ dV − ∇ · dV ≥ 0 ∀ΔV ⊂ V
dt dt θ θ

r
ba
ge ro or
V V V V

eS m
(5.107)

ci
f

ra
and localizing in (5.107), the local form of the second law of thermodynamics
C d P cs
b
a
or Clausius-Duhem equation is obtained.
i
an an n
y ha

Local form of the second law of thermodynamics


le
liv or ec

(Clausius-Duhem inequality)
(5.108)
ds  r  q 
M

.A

ds(i)
ρ = ρ − ρ −∇· ≥0 ∀x ∈ V, ∀t
dt dt θ θ
m

d
uu
e
X Th

er

Where, again, in (5.108) the sign


tin

= corresponds to reversible processes,


on

.O

> corresponds to irreversible processes, and


< indicates that the corresponding process is not feasible.
C

Equation (5.108) can be rewritten as follows.



q 1 ⎪

1 ⎪

∇· = ∇ · q − 2 q · ∇θ ⎪

θ θ θ ⎪

ds (i) ds r 1 1 ⇒ (5.109)
ρ =ρ −ρ + ∇ · q − 2 q · ∇θ ≥ 0 ⎪ ⎪

dt dt θ θ θ ⎪

  ⎪

not .(i) not . ⎭
=s =s

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Second Law of Thermodynamics. Entropy 239

. . r 1 1
s(i) = s − + ∇·q − 2 q · ∇θ ≥ 0 (5.110)
θ ρθ ρθ
 
.(i) .(i)
slocal scond

Then, a much stronger (more restrictive) formulation of the second law of


thermodynamics can be posed. This formulation postulates that the internally
. .(i)
generated entropy, s(i) , can be generated locally, slocal , or by heat conduction,
.(i)
scond , and that both contributions to the generation of entropy must be non-
negative.

rs
ee
s gin
Local internal generation of entropy
(Clausius-Planck inequality)
(5.111)

t d le En
.(i) . r 1
slocal = s − + ∇·q ≥ 0
θ ρθ

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
Internal generation of entropy by heat conduction
i
an an n

.(i) 1
y ha

(5.112)
scond = − q · ∇θ ≥ 0
ρθ 2
le
liv or ec
M

.A
m

d
uu
e

Remark 5.15. Equation (5.112) can be interpreted in the following


X Th

er
tin

manner: since the density, ρ, and the absolute temperature, θ , are


positive magnitudes, said equation can be written as
on

.O

q · ∇θ ≤ 0 ,
C

which establishes that the non-convective heat flux, q, and the tem-
perature gradient, ∇θ , are vectors that have opposite directions (their
dot product is negative). In other words, (5.112) is the mathemati-
cal expression of the experimentally verified fact that heat flows by
conduction from the hottest to the coldest parts in the medium (see
Figure 5.24), characterizing as non-feasible those processes in which
the contrary occurs.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
240 C HAPTER 5. BALANCE P RINCIPLES

rs
ee
Figure 5.25: Heat flux is opposed to the thermal gradient.

s gin
t d le En

r
ba
ge ro or
Remark 5.16. In the context of Fourier’s Law of heat conduction,

eS m
q = −K ∇θ (see Remark 5.9), expression (5.112) can be written as

ci
f

ra
!
C d P cs
b
a
q · ∇θ ≤ 0
i
=⇒ −K |∇θ |2 ≤ 0 =⇒ K ≥ 0
an an n

q = −K∇θ
y ha

le
liv or ec

which reveals that negative values of the thermal conductivity K lack


physical meaning.
M

.A
m

d
uu
e
X Th

er
tin

5.12.5 Alternative Forms of the Second Law of Thermodynamics


on

.O

Alternative expressions of the Clausius-Planck equation (5.111) in combination


with the local form of the energy balance equation (5.89) are often used in con-
C

tinuum mechanics.
©

• Clausius-Planck equation in terms of the specific internal energy


A common form of expressing the Clausius-Planck equation is doing so in terms
of the specific internal energy u (x,t) in (5.84). This expression is obtained using
the local spatial form of the energy balance equation (5.89),
du not . .
ρ = ρ u = σ : d + ρr − ∇ · q =⇒ ρr − ∇ · q = ρ u − σ : d , (5.113)
dt
and, replacing it in the Clausius-Planck equation (5.111),
.(i) . . .
ρθ slocal = ρθ s − (ρr − ∇ · q) = ρθ s − ρ u + σ : d ≥ 0 . (5.114)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Second Law of Thermodynamics. Entropy 241

Clausius-Planck equation in terms of the internal energy


. . (5.115)
−ρ (u − θ s) + σ : d ≥ 0

• Clausius-Planck equation in terms of the Helmholtz free energy


Another possibility is to express the Clausius-Planck equation in terms of the
(specific) Helmholtz free energy ψ (x,t), which is defined in terms of the internal
energy, the entropy and the temperature as

rs
ee
de f
ψ = u − sθ . (5.116)

s gin
Differentiating (5.116) with respect to time results in

t d le En
. . . . . . . .

r
ψ = u − sθ − sθ =⇒ u − θ s = ψ + sθ (5.117)

ba
ge ro or
eS m
ci
f
and, replacing (5.117) in (5.115), yields the Clausius-Planck equation in terms

ra
C d P cs
b
of the Helmholtz free energy,

a
i
. .
an an n

.(i) . .
ρθ slocal = −ρ (u − θ s) + σ : d = −ρ ψ + sθ + σ : d ≥ 0 .
y ha

(5.118)
le
liv or ec
M

.A

Clausius-Planck equation in terms of the free energy


. . (5.119)
−ρ ψ + sθ + σ : d ≥ 0
m

d
uu
e
X Th

.
er

For the infinitesimal strain case, d = ε (see Chapter 2, Remark 2.22), and re-
tin

placing in (5.119) results in


on

.O
C

Clausius-Planck equation (infinitesimal strain)


. .
©

. . (5.120)
−ρ ψ + sθ + σ : ε ≥ 0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
242 C HAPTER 5. BALANCE P RINCIPLES

5.13 Continuum Mechanics Equations. Constitutive


Equations
At this point it is convenient to summarize the set of (local) differential equations
provided by the balance principles.
1) Conservation of mass. Mass continuity equation.

dρ ⎪

+ ρ∇ · v = 0 ⎪

dt → 1 equation (5.121)
∂ vi ⎪

rs
dρ ⎪

+ρ =0 ⎭

ee
dt ∂ xi

s gin
t d le En
2) Balance of linear momentum. Cauchy’s equation.

r

ba
ge ro or
dv ⎪

eS m
∇ · σ + ρb = ρ ⎪

ci
dt
f
→ 3 equations

ra
(5.122)
C d P cs
∂ σ ji ⎪

b
dvi

a
+ ρbi = ρ i ∈ {1, 2, 3} ⎪

i
∂xj
an an n

dt
y ha

le
liv or ec

3) Balance of angular momentum. Symmetry of the stress tensor.


M

.A

!
σ = σT
m

→ 3 equations
d

(5.123)
uu

σ12 = σ21 ; σ13 = σ31 ; σ23 = σ32


e
X Th

er
tin
on

.O

4) Energy balance. First law of thermodynamics.



C



©

du
ρ = σ : d + (ρr − ∇ · q) ⎪ ⎬
dt
  → 1 equation (5.124)
du ∂ qi ⎪⎪

ρ = σi j di j + ρr − ⎭
dt ∂ xi

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Continuum Mechanics Equations. Constitutive Equations 243

5) Second law of thermodynamics. Clausius-Planck and heat flux inequalities.



. .
−ρ (u − θ s) + σ : d ≥ 0 ⎬
→ 1 restriction
−ρ (u − θ s) + σ d ≥ 0 ⎭
. .
ij ij

⎪ (5.125)
1 ⎪

− 2 q · ∇θ ≥ 0 ⎬
ρθ → 1 restriction
1 ∂θ ⎪

− 2 qi ≥0 ⎪

rs
ρθ ∂ xi

ee
These add up to a total of 8 partial differential equations (PDEs) and two re-

s gin
strictions. Counting the number of unknowns that intervene in these equations
results in25 ⎫

t d le En
ρ → 1 unknown ⎪ ⎪


r
v → 3 unknowns ⎪ ⎪

ba
ge ro or

eS m

σ → 9 unknowns ⎪

ci

⎬ f

ra
C d P cs
u → 1 unknown
b
19 unknowns

a


i

an an n

q → 3 unknowns ⎪ ⎪

y ha


θ → 1 unknown ⎪ ⎪


le

liv or ec

s → 1 unknown ⎭
M

.A

Therefore, it is obvious that additional equations are needed to solve the prob-
m

lem. These equations, which receive the generic name of constitutive equations
d
uu

and are specific to the material that constitutes the continuous medium, are
e
X Th

er
tin

6) Fourier’s law of heat conduction.



on

.O

q = −K ∇θ ⎪

C

∂θ → 3 equations (5.126)
i ∈ {1, 2, 3} ⎪

©

qi = −K
∂ xi

25The six components of the strain rate tensor d in (5.124) and (5.125) are not considered
unknowns because they are assumed to be implicitly calculable in terms of the velocity v by
means of the relation d (v) = ∇s v (see Chapter 2, Section 2.13.2).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
244 C HAPTER 5. BALANCE P RINCIPLES

7) Constitutive equations (per se)26 .



Thermo- ⎪

mechanical
σ , ε (v) , θ , μ ) = 0 i ∈ {1, ... , 6} → 6 equations
fi (σ
constitutive ⎪

equations
!
Entropy
constitutive s = s (εε (v) , θ , μ ) = 0 → 1 equation
equation
(5.127)
# $

rs
where μ = μ1 , ... , μ p are a set of new thermodynamic variables (p new

ee
unknowns) introduced by the thermo-mechanical constitutive equations.

s gin
8) Thermodynamic equations of state.

t d le En

 ⎪
Caloric ⎪

r
u = g (ρ, ε (v) , θ , μ ) ⎬

ba
ge ro or
eqn. of state

eS m
 → (1 + p) eqns.

ci

Kinetic f ⎪
i ∈ {1, 2, ... , p} ⎭

ra
Fi (ρ, θ , μ ) = 0
C d P cs
b
eqns. of state

a
i
(5.128)
an an n

There is now a set of (1 + p) equations and (1 + p) unknowns that, with the


y ha

adequate boundary conditions, constitute a mathematically well-defined prob-


le
liv or ec

lem.
M

.A
m

Remark 5.17. The mass continuity equation, Cauchy’s equation, the


d
uu

symmetry of the stress tensor, the energy balance and the inequalities
e
X Th

of the second law of thermodynamics (equations (5.121) to (5.125))


er
tin

are valid and general for all the continuous medium, regardless of the
material that constitutes the medium, and for any range of displace-
on

.O

ments and strains. Conversely, the constitutive equations (5.126) to


C

(5.128) are specific to the material or the type of continuous medium


©

being studied (solid, fluid, gas) and differentiate them from one an-
other.

26 The strains ε often intervene in the thermo-mechanical constitutive equations. However,


these are not considered as additional unknowns because they are assumed to be implicitly
calculable in terms of the equation of motion which, in turn, can be calculated by integration
of the velocity field, ε = ε (v) (see Chapters 1 and 2).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Continuum Mechanics Equations. Constitutive Equations 245

5.13.1 Uncoupled Thermo-Mechanical Problem


To solve the general problem in continuum mechanics, a system of partial dif-
ferential equations must be solved, which involve the (1 + p) equations and the
(1 + p) unknowns discussed in the previous section. However, under certain cir-
cumstances or hypotheses, the general problem can be decomposed into two
smaller problems (each of them involving a smaller number of equations and
unknowns), named mechanical problem and thermal problem, and that can be
solved independently (uncoupled) from one another.
For example, consider the temperature distribution θ (x,t) is known a priori,
or that it does not intervene in a relevant manner in the thermo-mechanical con-

rs
stitutive equations (5.127), and that, in addition, said constitutive equations do

ee
not involve new thermodynamic variables (μ μ = {0})./ In this case, the following

s gin
set of equations are considered 27

t d le En


dρ ⎪
(1 eqn) ⎪
Mass continuity ⎪
+ ρ∇ · v = 0 ⎪

r
equation: ⎪

ba
dt

ge ro or

eS m

ci
dv f (3 eqn) ⎪ → 10 equations ,

ra
∇ · σ + ρb = ρ
C d P cs
Cauchy’s equation:

b
a
dt ⎪


i

an an n



σ , ε (v)) = 0
fi (σ ⎪
y ha

(6 eqn) ⎪
Mechanical

constitutive equations: i ∈ {1, ... , 6}
le
liv or ec

(5.129)
M

.A

which involve the following unknowns.



m

ρ (x,t) → 1 unknown ⎪
d


uu
e

v (x,t) → 3 unknowns 19 unknowns (5.130)


X Th


er


tin

σ (x,t) → 6 unknowns
on

.O

The problem defined by equations (5.129) and (5.130) constitutes the so-
C

called mechanical problem, which involves the variables (5.130) (named me-
©

chanical variables) that, moreover, are the real interest in many engineering
problems.
The mechanical problem constitutes, in this case, a system of reduced differ-
ential equations, with respect to the general problem, and can be solved inde-
pendently of the rest of equations of said problem.

27 For simplicity, it is assumed that the symmetry of the stress tensor (5.123) is already
imposed. Then this equation is eliminated from the set of equations and the number of un-
knowns of σ is reduced from 9 to 6 components.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
246 C HAPTER 5. BALANCE P RINCIPLES

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 247

P ROBLEMS

Problem 5.1 – Justify whether the following statements are true or false.
a) The mass flux across a closed material surface is null only when the
motion is stationary.
b) The mass flux across a closed control surface is null when this flux is

rs
stationary.

ee
s gin
t d le En
Solution

r
a) The statement is false because a material surface is always constituted by

ba
ge ro or
eS m
the same particles and, therefore, cannot be crossed by any particle throughout

ci
f
its motion. For this reason, the mass flux across a material surface is always null,

ra
C d P cs
independently of the motion being stationary or not.
b
a
i
an an n

b) The statement is true because the application of the mass continuity equation
y ha

on a stationary flux implies


le
liv or ec



M

.A

∂ρ ⎪

Mass continuity equation =⇒ + ∇ · (ρv) = 0 ⎪ ⎬
∂t
m

=⇒ ∇ · (ρv) = 0 .
d



uu

∂ρ ⎪
e

Stationary flux =⇒ =0 ⎪

X Th

er

∂t
tin
on

Resulting, thus, what had to be proven,


.O

 
C

∇ · (ρv) = 0 =⇒ ∇ · (ρv) dV = ρv · n dS = 0 .
©

V ∂V

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
248 C HAPTER 5. BALANCE P RINCIPLES

Problem 5.2 – A water jet with cross-section S, pressure p and velocity v,


impacts perpendicularly on a disc as indicated in the figure below. Determine
the force F in steady-state regime that must be exerted on the disc for it to
remain in a fixed position (consider the atmospheric pressure is negligible).

rs
ee
s gin
t d le En

r
Solution

ba
ge ro or
eS m
ci
Taking into account the Reynolds Transport Theorem (5.39) and that the prob-
f

ra
lem is in steady-state regime, the forces acting on the fluid are
C d P cs
b
a
d   ∂  
i
∑ Fext/ f = ρv dV = (ρv) dV + ρv (n · v) dS = ρv (n · v) dS .
an an n

dt V V ∂t
y ha

∂V S
le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Note that the velocity vector of the fluid along the surfaces Slat−1 and Slat−3 is
perpendicular to the outward unit normal vector of the volume that encloses the
fluid, therefore, v · n = 0. The same happens in the walls of the disc.
The vectors v and n in sections S2 and S4 are not perpendicular but, because
there exists symmetry and v is perpendicular to F, they do not contribute com-
ponents to the horizontal forces. Therefore, the only forces acting on the fluid
are  
∑ ext/ f
F = ρv (n · v) dS = ρve (−e · ve) dS = −ρv2 Se .
∂V S

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 249

On the other hand, the external force, the pressure of the water jet and the atmo-
spheric pressure (which is negligible) also act on the fluid,

∑ Fext/ f = −Fe + atmospheric pressure forces + pSe = −Fe + pSe .


Equating both expressions and isolating the value of the module of the force F
finally results in
F = ρv2 S + pS .

rs
ee
Problem 5.3 – A volume flow rate Q circulates, in steady-state regime, through

s gin
a pipe from end A (with cross-section SA ) to end B (with cross-section SB < SA ).
The pipe is secured at point O by a rigid element P − O.

t d le En
Determine:

r
ba
ge ro or
eS m
a) The entry and exit velocities vA and vB in terms of the flow rate.

ci
f
b) The values of the angle θ that maximize and minimize the reaction force

ra
C d P cs
b
F at O, and the corresponding values of said reaction force.

a
i
c) The values of the angle θ that maximize and minimize the reaction mo-
an an n
y ha

ment M about O, and the corresponding values of said reaction mo-


le
ment.
liv or ec

d) The power W of the pump needed to provide the flow rate Q.


M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Hypotheses:
1) The water is a perfect fluid (σi j = −p δi j ) and incompressible.
2) The weight of the pipe and the water are negligible.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
250 C HAPTER 5. BALANCE P RINCIPLES

Solution
a) The incompressible character of water implies that the density is constant
for a same particle and, therefore, dρ/dt = 0. Introducing this into the mass
continuity equation (5.24), results in

∇·v = 0 ⇐⇒ ∇ · v dV = 0 ∀V . [1]
V

The adequate integration volume must now be defined. To this aim, a control
volume such that its boundary is a closed surface must be found (S = ∂V ) to be

rs
able to apply the Divergence Theorem,

ee
 

s gin
∇ · v dV = n · v dS ∀V [2]

t d le En
V ∂V

r
where n is the outward unit normal vector in the boundary of the volume V .

ba
ge ro or
eS m
Then, by means of [1] and [2], the conclusion is reached that the net outflow

ci
across the contour of the control volume is null, f

ra
C d P cs

b
a
i
n · v dS = 0 ∀V .
an an n
y ha

∂V
le
liv or ec

The volume the defined by the water contained inside the pipe between the cross-
M

.A

sections SA and SB is taken as control volume. Consider, in addition, the unit


vectors eA and eB perpendicular to said cross-sections, respectively, and in the
m

direction of the flow of water. Then, the following expression is deduced. Note
uu

that the extended integral on the boundary ∂V is applied only on cross-sections


e
X Th

SA and SB since n · v = 0 on the walls of the pipe, that is, n and v are perpendic-
er
tin

ular to one another.


on

.O

    
n · v dS = n · v dS + n · v dS = (−eA ) · vA eA dS + eB · vB eB dS = 0
C

∂V SA SB SA SB

=⇒ −vA SA + vB SB = 0 =⇒ vA SA = vB SB = Q

It is verified, thus, that the flow rate at the entrance and exit of the pipe are the
same,

Q Q
vA = ; vB = . [3]
SA SB

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 251

b) The balance of linear momentum equation (5.49) must be applied to find the
value of the force F,
  
d
R= ρb dV + t dS = ρv dV , [4]
dt
V ∂V V

where R is the total resultant of the forces acting on the fluid. On the other
hand, expanding the right-hand term in [4] by means of the Reynolds Transport
Theorem (5.39), yields
  

rs
d ∂
ρv dV = ρv dV + ρv (n · v) dS . [5]

ee
dt ∂t
V V ∂V

s gin
The problem is being solved for a steady-state regime, i.e., the local derivative

t d le En
of any property is null. In addition, the flow is known to exist solely through
sections SA and SB since n and v are perpendicular to one another on the walls

r
ba
ge ro or
of the pipe. Therefore, according to [4] and [5],

eS m
ci
f

ra
 
C d P cs
b
a
R= ρv (n · v) dS + ρv (n · v) dS =
i
an an n

SA SB
y ha

 
le
= ρvA eA (−eA · vA eA ) dS + ρvB eB (eB · vB eB ) dS
liv or ec
M

.A

SA SB
m

R = −ρv2A SA eA + ρv2B SB eB . [6]


uu
e
X Th

er
tin

Introducing [3] in [6] allows expressing the resultant force R in terms of Q,


on

.O

 
1 1
R = −ρQ − eA + eB .
2
C

SA SB
©

Now the different forces that compose R must be analyzed. According to the
statement of the problem, body forces can be neglected (b = 0). Therefore, only
surface forces must be taken into account, that is, the forces applied on the
boundary of the control volume (SA , SB and Slat , where this last one corresponds
to the lateral surface of the walls),

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
252 C HAPTER 5. BALANCE P RINCIPLES

     
R= ρb dV + t dS = t dS = t dS + t dS + t dS =
V ∂V ∂V SA SB Slat
 
= pA eA dS + pB (−eB ) dS + R p/ f .
SA SB

Here, R p/ f represents the forces exerted on the fluid by the walls of the pipe,
which initially are unknown but can be obtained using [6] as follows.
 

rs
R p/ f = R − pA eA dS − pB (−eB ) dS

ee
SA SB

s gin
R p/ f = −ρv2A SA eA + ρv2B SB eB − pA SA eA + pB SB eB

t d le En
   

r
R p/ f = − ρv2A + pA SA eA − ρv2B + pB SB eB [7]

ba
ge ro or
eS m
ci
f

ra
C d P cs
b
Introducing [3], R p/ f can be expressed in terms of Q,

a
i
an an n

   2 
Q2
y ha

Q
R p/ f =− ρ + pA SA eA − ρ + pB SB eB .
le
SA SB
liv or ec
M

.A

Now the relation between R p/ f and the unknown being sought, F, must be
found. To this aim, the action and reaction law is considered, and the pipe and
m

the rigid element P − O are regarded as a single body. Under these conditions,
d
uu

the force exerted by the fluid on the pipe is


e
X Th

er
tin

R f /p = −R p/ f .
on

.O

Since it is the only action on the body, and taking into account that the weight
C

of the pipe is negligible, this force must be compensated by an exterior action F


©

for the body to be in equilibrium.

R f /p + F = 0 =⇒ F = −R f /p = R p/ f
Introducing [7], the value of F is finally obtained as
   
F = − ρv2A + pA SA eA + ρv2B + pB SB eB .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 253

Using [3], the force F is expressed in terms of Q,


   2 
Q2 Q
F=− ρ + pA SA eA + ρ + pB SB eB . [8]
SA SB

There are two possible ways of obtaining the maximum and minimum of |F| in
terms of θ :
1) Determine the expression of |F| and search for its extremes by imposing
that its derivative is zero (this option not recommended).

rs
2) Direct method, in which the two vectors acting in the value of F are

ee
analyzed (this option developed below).

s gin
According to [7], the value of F depends on the positive scalar values FA and FB ,
which multiply the vectors (−eA ) and eB , respectively.

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu

The vector (−eA ) is fixed and does not depend on θ but eB does vary with θ . The
e
X Th

er

scalars FA and FB are constant values. Therefore, the maximum and minimum
tin

values of F will be obtained when FA and FB either completely add or subtract


on

one another, respectively. That is, when the vectors (−eA ) and eB are parallel to
.O

each other. Taking into account [3] and [8], the maximum and minimum values
C

are found to be:


©

− Minimum value of F
π
θ=
2
 
1 1
|F|min = ρQ 2
− + pB SB − pA SA
SB SA

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
254 C HAPTER 5. BALANCE P RINCIPLES

− Maximum value of F

θ=
2
 
1 1
|F|min = ρQ2 + + pB SB + pA SA
SB SA

c) The balance of angular momentum equation (5.57) must be applied to find


the moment M about point O,

rs
  
d

ee
Mliq = r × ρb dV + r × t dS = r × ρv dV , [9]
dt

s gin
V ∂V V

t d le En
where Mliq is the resultant moment of the moments acting on the fluid. On
the other hand, expanding the right-hand term in [9] by means of the Reynolds

r
ba
ge ro or
Transport Theorem (5.39), yields

eS m
ci
  
f

ra
d ∂
C d P cs
r × ρv dV = r × ρv dV + (r × ρv) (n · v) dS .
b
[10]

a
dt ∂t
i
an an n

V V ∂V
y ha

As in b), because the problem is in steady-state regime, the local derivative is


le
liv or ec

null. Again, n and v are perpendicular to one another on the walls of the pipe
M

and, thus, considering [9] and [10], results in the expression


.A

 
m

Mliq = (r × ρv) (n · v) dS + (r × ρv) (n · v) dS ,


d

[11]
uu
e

SA SB
X Th

er
tin

where the following must be taken into account:


on

.O

1. The solution to each integral can be determined considering the resultant


C

of the velocities in the middle point of each cross-section since the velocity
©

distributions are uniform and parallel in both cases.


2. For cross-section SA , the resultant of the velocity vector applied on the
center of the cross-section acts on point O and, therefore, does not generate
any moment because the cross product of the position vector at the center
of SA and the velocity vector are null.
3. For cross-section SB , vectors r and v belong to the plane of the paper and,
thus, their cross product has the direction of the vector (−ez ). In addition,
they are perpendicular to each other, so the module of their cross product
is the product of their modules.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 255

Applying these considerations to [11] yields



Mliq = R ρ vB (−ez ) (eB · vB eB ) dS
SB

Q2
Mliq = −ρ v2B R SB ez = −ρ R ez [12]
SB

The following step consists in studying the contributions of the body forces,
which in this case are null (b = 0), and of the surface forces.

rs
  

ee
Mliq = r × ρb dV + r × t dS = r × t dS =

s gin

V  ∂V  ∂V

t d le En
= r × t dS + r × t dS − r × t dS =

r
SA SB Slat


ba
ge ro or
eS m
= 0+ R pB ez dS + M p/ f = R pB SB ez + M p/ f ,

ci
f

ra
C d P cs
SB

b
a
i
an an n

where M p/ f is the moment exerted by the pipe on the fluid. To determine its
y ha

expression, [12] is used,


le
liv or ec

M p/ f = Mliq − R pB SB ez = −ρ v2B R SB ez − R pB SB ez ,
M

.A

 
  Q2
m

M p/ f = −R SB ρ v2B + pB ez = −R ρ + pB SB ez . [13]
d

SB
uu
e
X Th

er
tin

Introducing the action and reaction law will allow obtaining the moment exerted
by the fluid on the pipe,
on

.O

M p/ f = −M f /p .
C

Considering the pipe and the rigid element P − O as a single body in equilibrium
©

and neglecting the weight of the pipe,


M f /p + M = 0 =⇒ M = −M f /p = M p/ f .

Finally, the value of the moment M is obtained, using [13].


 
  Q2
M= −R SB ρv2B + pB ez = −R ρ + pB SB ez
SB

Note that this result does not depend on the angle θ and, therefore, its module
will have a constant value.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
256 C HAPTER 5. BALANCE P RINCIPLES

d) To determine the value of the power W needed to provide a volume flow rate
Q the balance of mechanical energy equation (5.73) is used.
 
d 1 2
W= ρv dV + σ : d dV [14]
dt 2
V V

The stress power in an incompressible perfect fluid is null,



σ : d dV = 0 .
V

rs
This is proven as follows.

ee
s gin
  
1 T
σ : d = −p 1 : d = −p Tr (d) = −p Tr l +l =

t d le En
2

r
⎡ ⎤

ba
ge ro or
∂ v x ∂ vx ∂ v x

eS m
ci
⎢ ⎥
f
⎢ ∂x ∂y ∂z ⎥

ra
⎢ ⎥
C d P cs

b ⎥

a
= −p Tr (ll ) = −p Tr ⎢ ∂ vy ∂ vy ∂ vy ⎥ =
i
⎢ ∂x ∂y ∂z ⎥
an an n

⎢ ⎥
y ha

⎣ ∂v ∂v ∂v ⎦
z z z
le
liv or ec

∂x ∂y ∂z
 
M

.A

∂ vx ∂ v y ∂ v z
= −p + + = −p ∇ · v = 0 ,
m

∂x ∂y ∂z
d
uu
e

where [1] has been applied in relation to the incompressibility condition, to con-
X Th

er
tin

clude that the divergence of the velocity is null.


Applying the Reynolds Transport Theorem (5.39) on the term of the material
on

.O

derivative of the kinetic energy in [14] results in


C

  
d 1 2 ∂ 1 2 1 2
©

W= ρv dV = ρv dV + ρv (n · v) dS .
dt 2 ∂t 2 2
V V ∂V

And, again, considering the problem is in steady-state regime and that n and v
are perpendicular to one another on the walls of the pipe, the expression of the
incoming power W is determined.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 257

 
1 2 1 2
W= ρv (n · v) dS + ρv (n · v) dS =
2 2
SA SB
 
1 2 1 2 1 1
= ρv (−vA ) dS + ρvB (vB ) dS = ρv3A SA + ρv3B SB
2 A 2 2 2
SA SB

Then, by means of [3], the final result is obtained.


 
1 1 1

rs
W = ρQ3 2
− 2
2 SB SA

ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
258 C HAPTER 5. BALANCE P RINCIPLES

E XERCISES

5.1 – Justify why the following statements are true.


a) In an incompressible flow, the volume flow rate across a control surface
is null.
b) In a steady-state flow, the mass flux across a closed control surface is
null.
c) In an incompressible fluid in steady-state regime, the density is uniform

rs
only when the density at the initial time is uniform.

ee
s gin
5.2 – The figure below shows the longitudinal cross-section of a square pipe.

t d le En
Water flows through this pipe, entering through section AE and exiting through
section CD. The exit section includes a floodgate BC that can rotate around

r
ba
ge ro or
hinge B and is maintained in vertical position by the action of force F.

eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Determine:
a) The exit velocity v2 in terms of the entrance velocity v1 (justify the ex-
pression used).
b) The resultant force and moment at point B of the actions exerted on the
fluid by the interior of the pipe.
c) The resultant force and moment at point B of the actions exerted by the
fluid on floodgate BC.
d) The value of the force F and the reactions the pipe exerts on flood-
gate BC.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 259

e) The power of the pump needed to maintain the flow.


Additional hypotheses:
1) Steady-state regime
2) Incompressible fluid
3) The pressures acting on the lateral walls of the pipe are assumed con-
stant and equal to the entrance pressure p.
4) The exit pressure is equal to the atmospheric pressure, which is negligi-
ble.

rs
5) Perfect fluid: σi j = −pδi j

ee
6) The weights of the fluid and the floodgate are negligible.

s gin
t d le En
5.3 – The figure below shows the longitudinal cross-section of a pump used to
inject an incompressible fluid, fitted with a retention valve OA whose weight,

r
ba
ge ro or
per unit of width (normal to the plane of the figure), is W . Consider a steady-

eS m
ci
state motion, driven by the velocity of the piston V and the internal uniform
f

ra
pressure P1 . The external uniform pressure is P2 .
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Determine:
a) The uniform velocities v1 and v2 in terms of V (justify the expression
used).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
260 C HAPTER 5. BALANCE P RINCIPLES

b) The resultant force, per unit of width, exerted by the fluid on the valve OA.
c) The resultant moment about O, per unit of width, exerted by the fluid on
the valve OA.
d) The value of W needed for the valve OA to maintain its position (as
shown in the figure) during the injection process.
Additional hypotheses:
1) The body forces of the fluid are negligible.
2) Perfect fluid: σi j = −pδi j

rs
Perform the analysis by linear meter.

ee
s gin
5.4 – A perfect and incompressible fluid flows through the pipe junction shown
in the figure below. The junction is held in place by a rigid element O − D.

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Determine:
a) The entrance velocities (vA and vB ) and the exit velocity (vC ) in terms
of the volume flow rate Q (justify the expression used).
b) The resultant force and moment at O of the actions exerted on the fluid
by the interior of the pipes in the junction.
c) The reaction force and moment at D of the rigid element.
d) The power W of the pump needed to provide the volume flow rates indi-
cated in the figure.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 261

Additional hypotheses:
1) The weights of the fluid and the pipes are negligible.

5.5 – The front and top cross-sections of an irrigation sprinkler are shown in
the figure below. A volume flow rate Q of water enters through section C at a
pressure P and exits through sections A and B at an atmospheric pressure Patm .
The flow is assumed to be in steady-state regime.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec

Determine:
M

.A

a) The entrance and exit velocities (justify the expression used).


m

b) The resultant force and moment at point O of the actions exerted on the
d
uu

fluid by the interior walls of the sprinkler.


e
X Th

er

c) The reaction that must be exerted on point O to avoid the sprinkler from
tin

moving in the vertical direction.


on

.O

d) The angular acceleration of the sprinkler’s rotation α. To this aim, as-


sume that I0 and I1 are, respectively, the central moments of inertia about
C

point O of the empty sprinkler and the sprinkler full of water.


©

e) The power needed to provide a volume flow rate 2Q, considering that
W ∗ is the power of the pump needed to provide a volume flow rate Q.
Additional hypotheses:
1) Incompressible fluid
2) Perfect fluid: σi j = −pδi j
3) The weights of the sprinkler and the water inside it are negligible.
4) SA = SB = S and SC = S∗
5) m = Iα

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
CH.6. LINEAR ELASTICITY
Multimedia Course on Continuum Mechanics
Overview
 Hypothesis of the Linear Elasticity Theory Lecture 1
 Linear Elastic Constitutive Equation
Lecture 2
 Generalized Hooke’s Law
 Elastic Potential Lecture 3
 Isotropic Linear Elasticity
 Isotropic Constitutive Elastic Constants Tensor
Lecture 4
 Lamé Parameters
 Isotropic Linear Elastic Constitutive Equation
 Young’s Modulus and Poisson’s Ratio
Lecture 5
 Inverse Isotropic Linear Elastic Constitutive Equation
 Spherical and Deviator Parts of Hooke’s Law Lecture 6
 Limits in Elastic Properties Lecture 7

2
Overview (cont’d)
 The Linear Elastic Problem
 Governing Equations Lecture 8
 Boundary Conditions
 The Quasi-Static Problem Lecture 9
 Solution
 Displacement Formulation
Lecture 10
 Stress Formulation Lecture 11
 Saint-Venant’s Principle
 Uniqueness of the solution
Lecture 12

 Linear Thermoelasticity
 Hypothesis of the Linear Elasticity Theory Lecture 13
 Linear Thermoelastic Constitutive Equation
 Inverse Constitutive Equation
Lecture 14
 Thermal Stress and Strain

3
Overview (cont’d)
 Thermal Analogies
 Solution to the linear thermoelastic problem Lecture 15
 1st Thermal Analogy
 2nd Thermal Analogy Lecture 16 Lecture 17

 Superposition Principle in Linear Thermoelasticity Lecture 18

 Hooke’s Law in Voigt Notation Lecture 19

4
6.1 Hypothesis of the Linear
Elasticity Theory
Ch.6. Linear Elasticity

5
Hypothesis of the Linear Elastic Model
 The simplifying hypothesis of the Theory of Linear Elasticity are:

1. ‘Infinitesimal strains and deformation’ framework

2. Existence of an unstrained and unstressed reference state

3. Isothermal, isentropic and adiabatic processes

8
Hypothesis of the Linear Elastic Model
1. ‘Infinitesimal strains and deformation’ framework
the displacements are infinitesimal:
 material and spatial configurations or coordinates are the same
≈0
x= X + u x≈X
 material and spatial descriptions of a property & material and spatial
differential operators are the same:
x=X γ ( x, t ) =
γ ( X, t ) =
Γ ( X, t ) =
Γ ( x, t )
∂ (•) ∂ (•)
= ∇ ( • )= ∇ ( • )
∂X ∂x
∂x
 the deformation gradient =
F ≈1 F ≈ 1, so the current spatial
∂X
density is approximated by the density at the reference configuration.
ρ 0 ρt F ≈ ρt
=
Thus, density is not an unknown variable in linear elastic problems.

9
Hypothesis of the Linear Elastic Model
1. ‘Infinitesimal strains and deformation’ framework
the displacement gradients are infinitesimal:
 The strain tensors in material and spatial configurations collapse
into the infinitesimal strain tensor.

E ( X, t ) ≈ e ( x, t ) =
ε ( x, t )

10
Hypothesis of the Linear Elastic Model
2. Existence of an unstrained and unstressed reference state
 It is assumed that there exists a reference unstrained and unstressed
neutral state, such that,
ε 0 ( x ) ε=
= ( x, t 0 ) 0
σ 0 ( x ) σ=
= ( x, t0 ) 0
 The reference state is usually assumed to correspond to the reference
configuration.

11
Hypothesis of the Linear Elastic Model
3. Isothermal and adiabatic (=isentropic) processes
 In an isothermal process the temperature remains constant.
θ ( x, t ) ≡ θ ( x, t0 ) ≡ θ 0 ( x ) ∀x

 In an isentropic process the entropy of the system remains constant.


ds
s ( X,=
t ) s ( X=
) = 0 s = 0
dt
 In an adiabatic process the net heat transfer entering into the body is
zero.

Q=
e ∫ρ
V
0 r dV − ∫ q ⋅ n dS=
∂V
0 ∀∆V ⊂ V

internal
heat conduction REMARK
from the exterior An isentropic process is an
sources
ρ0 r − ∇ ⋅ q= 0 ∀x ∀t idealized thermodynamic
process that is adiabatic,
isothermal and reversible.
12
6.2 Linear Elastic Constitutive
Equation
Ch.6. Linear Elasticity

13
Hooke’s Law
 R. Hooke observed in 1660 that, for relatively small deformations
of an object, the displacement or size of the deformation is
directly proportional to the deforming force or load.
σ

F= k ∆l F ∆l
=E E
A l
F ε
σ= E ε

 Hooke’s Law (for 1D problems) states that in an elastic


material strain is directly proportional to stress through the
elasticity modulus.

14
Generalized Hooke’s Law
 This proportionality is generalized for the multi-dimensional case
in the Theory of Linear Elasticity.
 σ ( x, t ) =
C(x) : ε ( x, t )
 Generalized
= σ Cijkl ε kl i, j ∈ {1, 2,3} Hooke’s Law
 ij
 It constitutes the constitutive equation of a linear elastic material.
 The 4th order tensor  is the constitutive elastic constants tensor:
 Has 34=81 components.
 Has the following symmetries, reducing the tensor to 21 independent
components:
minor Cijkl = C jikl REMARK
symmetries The current stress at a point
Cijkl = Cijlk depends only on the current strain
at the point, and not on the past
major Cijkl = C klij
symmetries history of strain states at the point.

15
Elastic Potential
 The internal energy balance equation for the (adiabatic) linear
elastic model is
global form stress power heat transfer rate
d d ρ0 u
dt V∫
ρ 0=
u dV ∫V = dt
dV ∫ σ : d dV + ∫ ( ρ 0 r − ∇ ⋅ q ) dV
V
= ε V
internal energy infinitesimal
strains REMARK
(Vt ≡ V ∀t )
local form The rate of strain tensor is related to the
d material derivative of the material strain
( ρ 0=
u ) σ : ε + ρ r − ∇ ⋅ q tensor through: E = FT ⋅ d ⋅ F
dt
 = ε and F = 1 .
In this case, E
Where:
 u is the specific internal energy (energy per unit mass).
 r is the specific heat generated by the internal sources.
 q is the heat conduction flux vector per unit surface.

16
Elastic Potential
 The stress power per unit of volume is an exact differential of the
internal energy density, û , or internal energy per unit of volume:
d duˆ ( x, t )  REMARK
( ρ 0 u )= = uˆ= σ : ε
dt  dt The symmetries of the
uˆ constitutive elastic
 Operating in indicial notation: constants tensor are used:
duˆ 1 Cijkl = C jikl
σ
=
 : 
ε 
=
ε σ 
=
ε C ε = ( ε ij Cijkl εkl + ε ij Cijkl εkl ) = minor
dt C:ε
ij
 ij ij ijkl kl
2    symmetries Cijkl = Cijlk
Cijkl ε kl i ↔k
j ↔l
major
= ( ε ij Cijkl εkl + εkl C klij εij ) = ( ε ij Cijkl εkl + εij Cijkl ε kl ) =
1 1 Cijkl = C klij
symmetries
2  2 
Cijkl d (ε C ε )
dt ij ijkl kl
1d
= εij Cijkl εkl duˆ d 1
= (ε : C : ε)
2 dt   dt dt 2
ε: C:ε
17
Elastic Potential
duˆ 1 d
= σ=: ε (ε : C : ε)
dt 2 dt
 Consequences:
1. Consider the time derivative of the internal energy in the whole
volume:
d d d ˆ
∫V dt ( ) ( ) U (t )
dt V∫ ∫ σ : ε dV
ˆ
u =
x , t dV ˆ
u =
x , t dV = stress
dt V power

 In elastic materials we talk about deformation energy because the stress


power is an exact differential.

REMARK
The stress power, in elastic materials is an
exact differential of the internal energy Uˆ .
Then, in elastic processes, we can talk of the
elastic energy Uˆ(t ) .
18
Elastic Potential
duˆ 1 d
= σ=: ε (ε : C : ε)
dt 2 dt
 Consequences:
2. Integrating the time derivative of the internal energy density,
1
=uˆ ( x, t ) ε ( x, t ) : C : ε ( x, t ) + a ( x )
2
and assuming that the density of the internal energy vanishes at the
neutral reference state, uˆ ( =
x, t0 ) 0 ∀x:
=0 1 1
uˆ ( ε )
1
ε ( x, t0 ) : C : ε ( x, t0 ) + a ( x ) =a (x) =0 ∀x = =ε
:C:ε σ (ε ) : ε
2 2 σ 2
 Due to thermodynamic reasons the internal energy is assumed
always positive 1
=uˆ ( ε ) ε : C : ε > 0 ∀ε ≠ 0
2
19
σ ( x, t ) = C( x ) : ε ( x, t )
Elastic Potential
 The internal energy density defines a potential for the stress
tensor, and is thus, named elastic potential. The stress tensor can
be computed as
= σ=
T
σ
∂uˆ ( ε( x, t )) ∂ 1 1 1 1 ∂û ( ε )
= ( ε : C : ε) = C : ε + ε : C= ( σ + σ)= σ σ=
∂ε ∂ε 2 2 2 2 ∂ε

 The constitutive elastic constants tensor can be obtained as the


second derivative of the internal energy density with respect to
the strain tensor field,

∂σ (ε) ∂ uˆ ( ε ) ∂ ( C : ε )
2
∂ 2uˆ ( ε )
= = = C Cijkl =
∂ε ∂ε ⊗ ∂ε ∂ε ∂εij ∂ε kl

20
6.3 Isotropic Linear Elasticity
Ch.6. Linear Elasticity

21
Isotropic Constitutive Elastic Constants
Tensor
 An isotropic elastic material must have the same elastic properties
(contained in C ) in all directions.
 All the components of C must be independent of the orientation of the
chosen (Cartesian) system C must be a (mathematically) isotropic
tensor.
C= λ 1 ⊗ 1 + 2 µ I

Cijkl =λδ ijδ kl + µ (δ ik δ jl + δ ilδ jk ) i, j , k , l ∈ {1, 2,3}

Where:
1
 [I ]ijkl
I is the 4th order unit tensor defined as= δ ik δ jl + δ ilδ jk 
2
 λ and µ are scalar constants known as Lamé‘s parameters or coefficients.
REMARK
The isotropy condition reduces the number of independent elastic constants from 21 to 2.

22
Isotropic Linear Elastic Constitutive
Equation
 Introducing the isotropic constitutive elastic constants tensor
C= λ 1 ⊗ 1 + 2 µ I into the generalized Hooke’s Law σ = C : ε ,
in index notation:
σ ij Cijkl=
= ε kl ( λδ δ
ij+ µ (δ ik δ jl + δ ilδ jk ) =
kl )
ε kl
= εij
1 1 
= λδ ij δ kl ε kl + 2 µ  δ ik δ jl ε kl + δ ilδ jk = ε kl  λTr (ε)δ ij + 2 µεij
2 2 
= εll = Tr (ε ) = ε=ji ε ij
1 1
= = εij + εij εij
2 2

 And the resulting constitutive equation is,


=σ λ Tr ( ε ) 1 + 2 µ ε Isotropic linear elastic
 constitutive equation.
λδ ij ε ll + 2µ ε ij i, j ∈ {1, 2,3}
σ ij = Hooke’s Law

23
Elastic Potential
 If the constitutive equation is,
=σ λ Tr ( ε ) 1 + 2 µ ε Isotropic linear elastic
constitutive equation.
λδ ij ε ll + 2µ ε ij i, j ∈ {1, 2,3}
σ ij = Hooke’s Law

Then, the internal energy density can be reduced to:


1 1 REMARK
=uˆ ( ε ) = σ :ε ( ε) : ε
λ Tr ( ε) 1 + 2 µ=
2 2    The internal energy density is an
=σ elastic potential of the stress tensor
1 1 as:
= λ Tr ( ε ) 1
 : ε + µε : ε
2=
2 Tr (ε) 2 ∂uˆ ( ε )
= σ= ( ε ) λ Tr ( ε ) 1 + 2µ ε
∂ε
1
= λ Tr 2 ( ε ) + µ ε : ε
2

24
Inversion of the Constitutive Equation
1. ε is isolated from the expression derived for Hooke’s Law
=σ λ Tr ( ε ) 1 + 2 µ ε =
ε
1

( σ − λ Tr ( ε ) 1 )

2. The trace of σ is obtained:


Tr ( λ Tr ( ε ) 1 + 2 µ=
Tr ( σ ) = ε ) λ Tr ( ε ) Tr (1 ) + 2 µ Tr ( ε ) =
( 3λ + 2µ ) Tr ( ε )
=3
3. The trace of ε is easily isolated:
1
Tr ( ε ) = Tr ( σ )
3λ + 2 µ

4. The expression in 3. is introduced into the one obtained in 1.


1  1  λ 1
=ε  σ − λ Tr ( σ ) 1  ε=− Tr ( σ ) 1 + σ
2µ  3λ + 2 µ  2 µ ( 3λ + 2 µ ) 2µ

25
Inverse Isotropic Linear Elastic
λ 1
Tr ( σ ) 1 +
Constitutive Equation ε=−
2 µ ( 3λ + 2 µ ) 2µ
σ

 The Lamé parameters in terms of E and ν :


µ ( 3λ + 2µ ) νE
E= λ=
λ+µ (1 +ν )(1 − 2ν )
λ E
ν= µ= G=
2 (λ + µ ) 2 (1 +ν )

 So the inverse const. eq. is re-written:


 ν 1 +ν
 ε = − Tr ( σ ) 1 + σ Inverse isotropic linear
E E
 elastic constitutive equation.
ε = ν 1 + ν
− σ ll δ ij + σ ij i, j ∈ {1, 2,3} Inverse Hooke’s Law.
 ij E E
εx =
1
E
(
σ x − ν (σ y + σ z ) ) γ xy =
G
1
τ xy

In engineering notation: εy =
1
E
(σ y − ν ( σ x + σ z ) ) 1
γ xz = τ xz
G
εz =
1
E
(
σ z − ν (σ x + σ y ) ) 1
γ yz = τ yz
G

26
Young’s Modulus and Poisson’s Ratio
 Young's modulus E is a measure of the stiffness of an elastic
material. It is given by the ratio of the uniaxial stress over the
uniaxial strain.
µ ( 3λ + 2µ )
E=
λ+µ

 Poisson's ratio ν is the ratio, when a solid is uniaxially stretched,


of the transverse strain (perpendicular to the applied stress), to the
axial strain (in the direction of the applied stress).

λ
ν=
2 (λ + µ )

27
Example
Consider an uniaxial traction test of an isotropic linear elastic material such
that:
σx > 0 E, ν
σ= σ= τ= τ= τ= 0 y

σx σx
y z xy xz yz

σx σx
x

Obtain the strains (in engineering notation) and comment on the results
obtained for a Poisson’s ratio of ν = 0 and ν = 0.5 .

28
εx =
1
(
σ x − ν (σ y + σ z ) ) γ xy =
1
τ xy
σx > 0 E G

Solution (σ y − ν ( σ x + σ z ) )
1 1
εy = γ xz = τ xz
σ=
y σ=z τ=
xy τ=
xz τ=
yz 0 E G
εz =
1
E
(
σ z − ν (σ x + σ y ) ) 1
γ yz = τ yz
G

For ν = 0 :
1
=εx = σx γ xy 0 1 There is no Poisson’s effect
E = σx
=εx
γ xy 0
ν
E and the transversal normal
εy =
− σx γ xz =
0 =ε y 0= γ xz 0 strains are zero.
E
=ε z 0= γ yz 0
ν
εz =
− σx γ yz =
0
E

For ν = 0.5 :
1 The volumetric deformation is
εx =
1
σx γ xy 0 =ε = σ γ 0
E
x
E
x xy
zero, tr ε = ε x + ε y + ε z = 0, the
0 material is incompressible
1
εy =

0.5
σx γ xz =
0 ε y =
− σx γ xz =
E 2E and the volume is preserved.
1
εz =

0.5
σx γ yz =
0 ε z =
− σx γ yz =
0
E 2 E

29
Spherical and deviatoric parts of
Hooke’s Law
 The stress tensor can be split into a spherical, or volumetric, part
and a deviatoric part:
1
σ= m1
sph : σ= Tr (σ ) 1
3 σ σ m1 + σ ′
=
σ ′ dev
= = σ σ − σ m1

 Similarly for the strain tensor:


1 1
=
ε sph =e1 Tr (ε ) 1 1
3 3 =ε e 1 + ε´
1 3
=ε′ dev
= ε ε − e1
3

30
Spherical and deviatoric parts of
Hooke’s Law
 Operating on the volumetric strain:
e = Tr ( ε )
ν 1 +ν
− Tr ( σ ) 1 +
ε= σ
E E

ν 1 +ν
− Tr ( σ ) Tr (1 ) +
e= Tr ( σ )
E E = 3σ m
=3

K : bulk modulus
3 (1 − 2ν ) E (volumetric strain modulus)
e= σm σm = e
E 3 (1 − 2ν ) def
2 E
K= λ+ µ=
3 3(1 − 2ν )
 The spherical parts of the stress and strain tensor are directly
related: σ = K e
m

31
Spherical and Deviator Parts of
Hooke’s Law
ν 1 +ν
 Introducing=
σ σ m 1 + σ ′ into ε= − Tr ( σ ) 1 + σ :
E E
ν 1 +ν
− Tr (σ m 1 + σ ′ ) 1 +
ε= (σ m 1 + σ ′ )
E E
ν ν =0 1 +ν 1 +ν  1 +ν 3ν  1 +ν
− σ mTr (1 ) 1 − Tr (σ ′ ) 1 + σ m
= 1+ σ′ =  − σ
 m 1 + σ′
E =3 E E E  E E  E
E
Taking into account that σm = e :
3 (1 − 2ν ) Comparing this
 1 − 2ν 1 E 1 +ν 1 1 +ν with the expression
ε=
  e 1 + σ ′ = e 1 + σ′
 E  3 (1 − 2ν ) E 3 E =ε
1
e 1 + ε´
3
1 +ν 1 1 1 +ν
= = ε´= σ′
E 2µ 2G E
 The deviatoric parts of the stress and strain tensor are related
component by component:
= σ ′ 2G ε′= σ ij′ 2Gε ij′ i, j ∈ {1, 2,3}

32
Spherical and deviatoric parts of
Hooke’s Law
 The spherical and deviatoric parts of the strain tensor are
directly proportional to the spherical and deviatoric parts
(component by component) respectively, of the stress tensor:

σm = K e σ ij′ = 2G ε ij′

33
Elastic Potential
 The internal energy density û ( ε ) defines a potential for the stress
tensor and is, thus, an elastic potential:
REMARK
1
uˆ ( ε ) = ε : C : ε ∂uˆ ( ε ) The constitutive elastic constants
=σ = :ε
2 ∂ε tensor C is positive definite due
to thermodynamic considerations.
 Plotting û ( ε ) vs. ε :
There is a minimum for ε = 0 :

∂uˆ ( ε )
= (C : ε) =0
∂ε ε =0 ε =0

∂ 2uˆ ( ε ) ∂ 2uˆ ( ε )
= = C= C

=
ε ⊗ ∂ ε ∂
ε 0=
ε ⊗ ∂ ε ε 0
ε =0

34
Elastic Potential
 The elastic potential can be written as a function of the spherical
and deviatoric parts of the strain tensor:
ε : C : ε = (C : ε) : ε = σ : ε
1 1 1
uˆ ( ε )
= =ε : C :=
ε σ:ε λTr ( ε ) 1 + 2 µε  : ε =
2 2 2

1  1 
=  e 1 + ε´  
: e 1 + ε ´ =
1 = Tr ( ε ) = e 1 = e 2
 3   3 
= λTr ( ε ) 1 : ε + µε= :ε λTr ( ε ) + µ ε : ε
2
1 2 2
2 2 = e 1 : 1 + e1 : ε ´ + ε´= : ε´
9 3
3 Tr ( ε′ ) =0
K
1 1 1 2  1
uˆ ( ε ) = λ e 2 + µ e 2 + μ ε´: ε´=  λ + µ  e 2 + μ ε´: ε´ = e 2 + ε´: ε´
2 3 2 3  3
Elastic potential in terms of the
1
uˆ ( ε ) = K e + µ ε´: ε´≥ 0 spherical and deviatoric parts
2

2 of the strains.
35
Limits in the Elastic Properties
 The derived expression must hold true for any deformation
process: 1
uˆ ( ε ) = : K e 2 + µ ε´ : ε´≥ 0
2
 Consider now the following particular cases of isotropic linear
elastic material:
 Pure spherical deformation process
1
ε (1) = e 1 1
3 uˆ (1)
= K e2 ≥ 0 K >0 bulk modulus
2
ε′(1) = 0
 Pure deviatoric deformation process
ε ( 2 ) = ε′ Lamé’s second
=
uˆ( 2)
µ ε´: ε´≥ 0 µ >0
e( 2 ) = 0 parameter

REMARK
ε′=
: ε′ ε ij ε ij ≥ 0
36
Limits in the Elastic Properties
 K and µ are related to E and ν through:

E E
=K >0 µ= G= >0
3 (1 − 2ν ) 2 (1 +ν )
REMARK
In rare cases, a material can
 Poisson’s ratio has a non-negative value, have a negative Poisson’s ratio.
E Such materials are named
>0
2 (1 +ν ) E≥0 Young’s auxetic materials.
modulus
ν ≥0
 Therefore,
E
>0
3 (1 − 2ν ) 0 ≤ν ≤
1
Poisson’s ratio
E≥0 2

37
6.4 The Linear Elastic Problem
Ch.6. Linear Elasticity

38
Introduction
 The linear elastic solid is subjected to body forces and prescribed
tractions:
Initial actions:
b ( x, 0 )
t =0
t ( x, 0 )

b ( x, t )
Actions
through time: t ( x, t )

 The Linear Elastic problem is the set of equations that allow


obtaining the evolution through time of the corresponding
displacements u ( x,t ) , strains ε ( x,t ) and stresses σ ( x,t ) .

39
Governing Equations
 The Linear Elastic Problem is governed by the equations:
1. Cauchy’s Equation of Motion.
Linear Momentum Balance Equation.
∂ 2 u ( x, t )
∇ ⋅ σ ( x, t ) + ρ 0 b ( x, t ) =
ρ0
∂ t2

2. Constitutive Equation.
Isotropic Linear Elastic Constitutive Equation. This is a PDE system of
σ ( x, t ) λTr ( ε ) 1 + 2 µ ε
= 15 eqns -15 unknowns:
u ( x,t ) 3 unknowns
3. Geometrical Equation. ε ( x,t ) 6 unknowns
Kinematic Compatibility. σ ( x,t ) 6 unknowns
ε ( x, t )= ∇ S u ( x, t )=
1 Which must be solved in
(u ⊗ ∇ + ∇ ⊗ u )
2 the R 3 × R +space.

40
Boundary Conditions
 Boundary conditions in space
 Affect the spatial arguments of the unknowns
 Are applied on the boundary Γ of the solid,
which is divided into three parts:
 Prescribed displacements on Γ u :
u(x, t ) = u* (x, t )
 ∀x ∈ Γu ∀t
=ui ( x, t ) ui ( x, t ) i ∈ {1, 2,3}
*

 Prescribed tractions on Γσ : Γu  Γσ  Γuσ = Γ ≡ ∂V


σ ( x, t ) ⋅ n = t * ( x, t ) Γu  Γσ =Γu  Γuσ =Γuσ  Γσ ={0}
/
 ∀x ∈ Γσ ∀t
σ ij ( x,=
t ) ⋅ n j t *j ( x, t ) i ∈ {1, 2,3}

 Prescribed displacements and stresses on Γ uσ :


ui ( x, t ) = ui* ( x, t )

σ ( ) ⋅ = *
( )
( i, j, k ∈ {1, 2,3} i ≠ j ) ∀x ∈ Γuσ ∀t
 jk x , t nk t j x, t

41
Boundary Conditions

42
Boundary Conditions
 Boundary conditions in time. INTIAL CONDITIONS.
 Affect the time argument of the unknowns.
 Generally, they are the known values at t = 0 :
 Initial displacements:
u ( x, 0=
) 0 ∀x ∈ V

 Initial velocity:
∂u ( x, t ) not
= u (= x, 0 ) v 0 ( x ) ∀x ∈ V
∂t t =0

43
The Linear Elastic Problem
 Find the displacements u ( x,t ) , strains ε ( x,t ) and stresses σ ( x,t )
such that
∂ 2 u ( x, t )
∇ ⋅ σ ( x, t ) + ρ 0 b ( x, t ) =
ρ0 Cauchy’s Equation of Motion
∂ t2
σ ( x, t ) λTr ( ε ) 1 + 2 µ ε
= Constitutive Equation
1
ε ( x, t )= ∇ S u ( x, t )= (u ⊗ ∇ + ∇ ⊗ u ) Geometric Equation
2

Γu : u =
u*
Boundary conditions in space
σ ⋅n
Γσ : t * =

u ( x, 0 ) = 0
Initial conditions (Boundary conditions in time)
u ( x, 0 ) = v 0

44
Actions and Responses
 The linear elastic problem can be viewed as a system of actions or data
inserted into a mathematical model made up of the EDP’s and
boundary conditions , which gives a response (or solution) in
displacements, strains and stresses.
b ( x, t )
u ( x, t )
t* ( x, t ) Mathematical
model ε ( x, t )
u* ( x, t )
EDPs+BCs σ ( x, t )
v0 ( x )
not
not RESPONSES = R ( x,t )
ACTIONS = A ( x,t )
 Generally, actions and responses depend on time. In these cases, the
problem is a dynamic problem, integrated in R3 × R + .
 In certain cases, the integration space is reduced to R3 . The problem is
termed quasi-static.
45
The Quasi-Static Problem
 A problem is said to be quasi-static if the acceleration term can
be considered to be negligible.
∂ 2 u ( x, t )
=a ≈0
∂t 2

 This hypothesis is acceptable if actions are applied slowly. Then,


∂ 2 u ( x, t )
∂ A / ∂t ≈ 0
2 2
∂ R / ∂t ≈ 0
2 2
≈0
∂ t2

46
The Quasi-Static Problem
 Find the displacements u ( x,t ) , strains ε ( x,t ) and stresses σ ( x,t )
such that ∂ u ( x,t ) 2

ρ0 ≈0
∂ t2
∇ ⋅ σ ( x, t ) + ρ0b ( x, t ) =
0 Equilibrium Equation

σ ( x, t ) λTr ( ε ) 1 + 2 µ ε
= Constitutive Equation
1
ε ( x, t )= ∇ S u ( x, t )= (u ⊗ ∇ + ∇ ⊗ u ) Geometric Equation
2

Γu : u =
u*
Boundary Conditions in Space
σ ⋅n
Γσ : t * =

u ( x, 0 ) = 0
Initial Conditions
u ( x, 0 ) = v 0

47
The Quasi-Static Problem
 The quasi-static linear elastic problem does not involve time
derivatives.
 Now the time variable plays the role of a loading descriptor: it describes the
evolution of the actions.

b ( x, λ ) u ( x, λ )
Mathematical
t* ( x, λ ) model ε ( x, λ )
u* ( x, λ ) EDPs+BCs σ ( x, λ )
not not
ACTIONS = A ( x, λ ) RESPONSES = R ( x, λ )

 For each value of the actions A ( x, λ * ) -characterized by a fixed value λ *- a


response R ( x, λ * ) is obtained.
 Varying λ * , a family of actions and its corresponding family of responses
are obtained.

48
Example
Consider the typical material strength problem where a cantilever beam is
subjected to a force F ( t ) at it’s tip.
For a quasi-static problem,

The response is δ ( t ) = δλ ((t )) , so for every time instant, it only depends on


the corresponding value λ ( t ) .

49
Solution of the Linear Elastic Problem
 To solve the isotropic linear elastic problem posed, two approaches can
be used:
 Displacement formulation - Navier Equations
Eliminate σ ( x,t ) and ε ( x,t ) from the general system of equations. This
generates a system of 3 eqns. for the 3 unknown components of u ( x,t ) .
 Useful with displacement BCs.
 Avoids compatibility equations.
 Mostly used in 3D problems.
 Basis of most of the numerical methods.

 Stress formulation - Beltrami-Michell Equations.


Eliminates u ( x,t ) and ε ( x,t ) from the general system of equations. This
generates a system of 6 eqns. for the 6 unknown components of σ ( x,t ) .
 Effective with boundary conditions given in stresses.
 Must work with compatibility equations.
 Mostly used in 2D problems.
 Can only be used in the quasi-static problem.

50
Displacement formulation
∂ 2 u ( x, t )
∇ ⋅ σ ( x, t ) + ρ 0 b ( x, t ) =
ρ0 Cauchy’s Equation of Motion
∂ t2
σ ( x, t ) λTr ( ε ) 1 + 2 µ ε
= Constitutive Equation
1
ε ( x, t )= ∇ S u ( x, t )= (u ⊗ ∇ + ∇ ⊗ u ) Geometric Equation
2

Γu : u =
u*
Boundary Conditions in Space
σ ⋅n
Γσ : t * =

u ( x, 0 ) = 0
Initial Conditions
u ( x, 0 ) = v 0

The aim is to reduce this system to a system with u ( x,t ) as the only unknowns.
Once these are obtained, ε ( x,t ) and σ ( x,t ) will be found through substitution.

51
Displacement formulation
 Introduce the Constitutive Equation into Cauchy’s Equation of
motion:
σ ( x, t ) λTr ( ε ) 1 + 2 µ ε
=
∂ 2u
λ ∇ ⋅ [Tr (ε) 1] + 2µ∇ ⋅ ε + ρ0b =ρ0 2
∂ 2 u ( x, t ) ∂t
∇ ⋅ σ ( x, t ) + ρ 0 b ( x, t ) =
ρ0
∂ t2
Consider the following identities:
= ( ∇ ( ∇ ⋅u ) )i

∂ ∂  ∂uk  ∂  ∂uk  ∂
∇ ⋅ ( Tr ( ε) 1)=
i
∂x j
( )
ε11 δ ij=
∂x j  ∂x δ ij=
 ∂x  ∂x =
∂xi
( ∇ ⋅ u=)
 k  i  k 

= ∇⋅u
=∇ ( ∇ ⋅ u ) i i ∈ {1, 2,3}

∇ ⋅ (Tr ( ε ) 1 ) =
∇ ( ∇ ⋅u )

52
Displacement formulation
 Introduce the Constitutive Equation into Cauchy’s Equation of
motion:
=σ ( x, t ) λTr ( ε ) 1 + 2 µ ε
∂ 2u
λ ∇ ⋅ [Tr (ε) 1] + 2µ∇ ⋅ ε + ρ0b =ρ0 2
∂ 2 u ( x, t ) ∂t
∇ ⋅ σ ( x, t ) + ρ 0 b ( x, t ) =
ρ0
∂ t2
Consider the following identities:
( )

= ∇2 u = ( ∇ ( ∇ ⋅ u ) )i
i
∂ε ij
( ∇ ⋅ ε )i = =
∂  1  ∂ui ∂u j   1 ∂ 2ui
  +   = +
1 ∂
∂x j ∂x j  2  ∂x j ∂xi   2 ∂x j ∂x j 2 ∂xi
 ∂u j

 1 2
=
 ∂x j 2 ( )
∇ u +
i
1 ∂
2 ∂xi
( ∇ ⋅ u )=

1 1  = ∇⋅u
=  ∇ 2u + ∇ ( ∇ ⋅ u )  i ∈ {1, 2,3}
2 2 i
1 1
∇=
⋅ε ∇ (∇ ⋅ u) + ∇ 2u
2 2

53
Displacement formulation
 Introduce the Constitutive Equation into Cauchy’s Equation of
Movement:
σ ( x, t ) λTr ( ε ) 1 + 2 µ ε
=
∂ 2u
λ ∇ ⋅ [Tr (ε) 1] + 2µ∇ ⋅ ε + ρ0b =ρ0 2
∂ 2 u ( x, t ) ∂t
∇ ⋅ σ ( x, t ) + ρ 0 b ( x, t ) =
ρ0
∂ t2
 Replacing the identities:
∇ ⋅ (Tr ( ε ) 1 ) =
∇ ( ∇ ⋅u ) ∇=
⋅ε
1 1
∇ (∇ ⋅ u) + ∇ 2u
2 2
1 1 2  ∂ 2u 2nd order
 Then, λ ∇ ( ∇ ⋅ u ) + 2 µ  ∇ (∇ ⋅ u) + ∇ u  + ρ0b =
ρ0 2 PDE system
 2 2  ∂t
 ∂ 2u
( λ + µ ) ∇ ( ∇ ⋅ u ) + µ∇ u + ρ0b = ρ0 2
2
 The Navier Equations ∂t

are obtained: ( λ + µ ) u j , ji + µ ui , jj + ρ
= 0 bi ρ0ui i ∈ {1, 2,3}

54
Displacement formulation
 The boundary conditions are also rewritten in terms of u ( x,t ) :
σ ( x, t ) λTr ( ε ) 1 + 2 µ ε
= = ∇⋅u
=t* λ (Tr ( ε ) ) n + 2 µ ε ⋅ n
1
t = σ ⋅n
* = ∇ S u= (u ⊗ ∇ + ∇ ⊗ u )
2

t=
*
λ (∇ ⋅ u ) n + µ (u ⊗ ∇ + ∇ ⊗ u ) ⋅ n
 The BCs are now:

u = u*
 on Γu
=ui ui i ∈ {1, 2,3}
*

λ ( ∇ ⋅ u ) n + µ ( u ⊗ ∇ + ∇ ⊗ u ) ⋅ n = t* REMARK
 on Γσ The initial conditions
λ uk .k ni + µ ( ui , j n j + u j ,i n j ) =
ti* i ∈ {1, 2,3}
remain the same.

55
Displacement formulation
 Navier equations in a cylindrical coordinate system:
∂e 2G ∂ω z ∂ωθ ∂ 2 ur  x = r cos θ
( λ + 2G ) − + 2G + ρ br = ρ 2 
∂r r ∂θ ∂z ∂t x( r , θ , z ) ≡  y = r sin θ
1 ∂e ∂ωr ∂ω z ∂ 2uθ z = z
( λ + 2G ) − 2G + 2G + ρ bθ =ρ 2 
r ∂θ ∂z ∂r ∂t
∂e 2G ∂ 2G ∂ωr ∂ 2u z
( λ + 2G ) − ( rωθ ) + + ρ bz =
ρ 2
∂z r ∂r r ∂θ ∂t

1  1 ∂u z ∂uθ 
ωr = −Ωθ z =  − 
2  r ∂θ ∂z 
1  ∂ ur ∂ u z  dV = r dθ dr dz
Where: ωθ = −Ω zr =  − 
2 ∂z ∂r 
1  1 ∂ ( ruθ ) 1 ∂ur  1 ∂ 1 ∂u ∂u
ω z = −Ω rθ = 
2  r ∂r
− 
r ∂θ 
=
e ( rur ) + θ + z
r ∂r r ∂θ ∂z

56
Displacement formulation
 Navier equations in a spherical coordinate system:
∂e 2G ∂ 2G ∂ωθ ∂ 2 ur
( λ + 2G ) −
∂r r sin θ ∂θ
(ωϕ sin θ ) + r sin θ ∂ϕ + ρ br =
ρ 2
∂t  x = r sin θ cos ϕ

=x x ( r ,θ , ϕ ) =
≡  y r sin θ sin ϕ
( λ + 2G ) ∂e − 2G ∂ωr ∂ ∂u
2

r ∂θ
+
2G
r sin θ ∂ϕ r sin θ ∂r
( )
rωϕ sin θ + ρ bθ =
ρ 2θ
∂t
 z = r cos θ

( λ + 2G ) ∂e − 2G ∂ rω + 2G ∂ωr + ρ b = ∂ 2uϕ
( θ) ϕ ρ 2 dV = r 2 sin θ dr dθ dϕ
r sin θ ∂ϕ r ∂r r ∂θ ∂t
1 1 ∂ 1 ∂uθ 
ωr = −Ωθϕ = 
2  r sin θ ∂θ
( uϕ sin θ ) − 
r sin θ ∂φ 
Where: 1  1 ∂ur 1 ∂ ( ruϕ ) 
ωθ = −Ωϕ r =  − 
2  r sin θ ∂φ r ∂r 
 
11 ∂ 1 ∂u
ωϕ = −Ω rθ =  ( ruθ ) − r 
2  r ∂r r ∂θ 
∂ 2 ∂ ∂ 
 ∂r ( r ur sin θ ) + ∂θ ( ruθ sin θ ) + ∂ϕ ( ruϕ ) 
1
=e
r sin θ
2
 
57
Stress formulation
Equilibrium Equation
∇ ⋅ σ ( x, t ) + ρ0b ( x, t ) =
0
(Quasi-static problem)
ν 1 +ν
ε ( x,t ) =
− Tr ( σ ) 1 + σ Inverse Constitutive Equation
E E
1
ε ( x, t )= ∇ S u ( x, t )= (u ⊗ ∇ + ∇ ⊗ u ) Geometric Equation
2

Γu : u =
u*
Boundary Conditions in Space
σ ⋅n
Γσ : t * =

The aim is to reduce this system to a system with σ ( x,t ) as the only unknowns.
Once these are obtained, ε ( x,t ) will be found through substitution and u ( x,t ) by
integrating the geometric equations.
REMARK
For the quasi-static problem, the time variable plays the role of a loading factor.

58
Stress formulation
 Taking the geometric equation and, through successive derivations,
the displacements are eliminated:
∂ 2ε ij ∂ 2ε kl ∂ 2ε ik ∂ 2ε jl Compatibility Equations
+ − − = 0 i, j , k , l ∈ {1, 2,3} (seen in Ch.3.)
∂xk ∂xl ∂xi ∂x j ∂x j ∂xl ∂xi ∂xk

 Introducing the inverse constitutive equation into the compatibility


equations and using the equilibrium equation:
ν 1 +ν
εij =− σ ppδ ij + σij
E E
∂σ ij
+ ρ 0b j =
0
∂xi
2nd order
 The Beltrami-Michell Equations are obtained: PDE system
1 ∂ 2σ kk ν ∂ ( ρ0bk ) ∂ ( ρ0bi ) ∂ ( ρ0b j )
∇ σ ij +
2
=
− δ ij − − i, j ∈ {1, 2,3}
1 +ν ∂xi ∂x j 1 −ν ∂xk ∂x j ∂xi

59
Stress formulation
 The boundary conditions are:
 Equilibrium Equations: ∇ ⋅ σ + ρ0b =0
This is a 1st order PDE system, so they can act as boundary conditions of
the (2nd order PDE system of the) Beltrami-Michell Equations

 Prescribed stresses on : Γσ =
σ ⋅ n t* on Γσ

60
Stress formulation
 Once the stress field is known, the strain field is found by substitution.
ν 1 +ν
− Tr ( σ ) 1 +
ε ( x,t ) = σ
E E
 The calculation, after, of the displacement field requires that the
geometric equations be integrated with the prescribed displacements
on Γu: 1
ε=
( x) ( u ( x) ⊗ ∇ + ∇ ⊗ u ( x) ) x ∈ V
2
=
u ( x ) u* ( x ) ∀x ∈ Γu

REMARK
This need to integrate the second system is a considerable disadvantage with
respect to the displacement formulation when using numerical methods to solve
the lineal elastic problem.

61
Saint-Venant’s Principle
 From A. E. H. Love's Treatise on the mathematical theory of elasticity:
“According to the principle, the strains that are produced in a body by the application,
to a small part of its surface, of a system of forces statically equivalent to zero force
and zero couple, are of negligible magnitude at distances which are large compared
with the linear dimensions of the part.” REMARK
This principle does not have a
 Expressed in another way: rigorous mathematical proof.
“The difference between the stresses caused by statically equivalent load systems is
insignificant at distances greater than the largest dimension of the area over which
the loads are acting.”

u (I) ( x P , t ) ≈ u (II) ( x P , t )
ε (I) ( x P , t ) ≈ ε (II) ( x P , t ) ∀P | δ >> 
σ (I) ( x P , t ) ≈ σ (II) ( x P , t )

62
Saint-Venant’s Principle
 Saint Venant’s Principle is often used in strength of materials.
 It is useful to introduce the concept of stress:
The exact solution of this problem is very complicated.

This load system is statically equivalent to load system (I).


The solution of this problem is very simple.

Saint Venant’s Principle allows approximating solution (I) by solution (II) at a


far enough distance from the ends of the beam.
63
Uniqueness of the solution
 The solution of the lineal elastic problem is unique:
 It is unique in strains and stresses.
 It is unique in displacements assuming that appropriate boundary
conditions hold in order to avoid rigid body motions.

 This can be proven by Reductio ad absurdum ("reduction to the


absurd"), as shown in pp. 189-193 of the course book.
 This proof is valid for lineal elasticity in infinitesimal strains.
 The constitutive tensor C is used, so proof is not only valid for isotropic
problems but also for orthotropic and anisotropic ones.

64
6.5 Linear Thermoelasticity
Ch.6. Linear Elasticity

65
Hypothesis of the Linear Thermo-elastic
Model
 The simplifying hypothesis of the Theory of Linear Thermo-
elasticity are:
1. Infinitesimal strains and deformation framework
 Both the displacements and their gradients are infinitesimal.

2. Existence of an unstrained and unstressed reference state


 The reference state is usually assumed to correspond to the reference
configuration. =ε ( x ) ε=
0 ( x, t ) 0
0

σ 0 ( x ) σ=
= ( x, t0 ) 0
3. Isentropic and adiabatic processes – no longer isothermal !!!
 Isentropic: entropy of the system remains constant
 Adiabatic: deformation occurs without heat transfer

66
Hypothesis of the Linear Thermo-Elastic
Model
3. (Hypothesis of isothermal process is removed)
 The process is no longer isothermal so the temperature changes
throughout time: not
θ ( x, t ) ≠ θ ( x, 0 ) = θ0
∂θ ( x, t )
θ ( x, t )
= ≠0
∂t
We will assume the temperature field is known.

 But the process is still isentropic and adiabatic:


s ( t ) ≡ cnt s = 0

Q=e ∫ ρ r dV − ∫ q ⋅ n dS=
∂V
0 ∀∆V ⊂ V
internal V
heat conduction
sources from the exterior
ρ r − ∇ ⋅ q= 0 ∀x ∀t
67
Generalized Hooke’s Law
 The Generalized Hooke’s Law becomes:
σ ( x, t ) C : ε ( x, t ) − β (θ=
 = − θ 0 ) C : ε ( x, t ) − β ∆θ Generalized Hooke’s Law for

σ ij= Cijkl ε kl − βij (θ − θ 0 ) i, j ∈ {1, 2,3} linear thermoelastic problems

Where
 is the elastic constitutive tensor.
 θ ( x,t ) is the absolute temperature field.
 θ 0 = θ ( x,t0 ) is the temperature at the reference state.
 β is the tensor of thermal properties or constitutive thermal
constants tensor.
 It is a positive semi-definite symmetric second-order tensor.
REMARK
A symmetric second-order tensor A is positive semi-definite
when zT·A·z > 0 for every non-zero column vector z.
68
Isotropic Constitutive Constants Tensors
 An isotropic thermoelastic material must have the same elastic
and thermal properties in all directions:
  must be a (mathematically) isotropic 4th order tensor:
C= λ 1 ⊗ 1 + 2 µ I

Cijkl =λδ ijδ kl + µ (δ ik δ jl + δ ilδ jk ) i, j , k .l ∈ {1, 2,3}
Where:
1
th =
 I is the 4 order symmetric unit tensor defined as [ I ] δ ik δ jl + δ ilδ jk 
ijkl
2
 λ and µ are the Lamé parameters or coefficients.

β is a (mathematically) isotropic 2nd order tensor:


β = β 1

= βij β δ ij i, j ∈ {1,2,3}
Where:
 β is a scalar thermal constant parameter.

69
Isotropic Linear Thermoelastic
Constitutive Equation
Introducing the isotropic constitutive constants tensors β = β 1 and
C= λ 1 ⊗ 1 + 2 µ I into the generalized Hooke’s Law, σ = C : ε − β (θ − θ 0 )
(in indicial notation)
σ=
ij Cijkl ε kl −βij (θ − θ 0=) ( λδ δ
ij kl )
+ µ (δ ik δ jl + δ ilδ jk ) ε kl − β (θ − θ 0 ) δ ij=
= ε ij
1 1 
= λδ ij δ kl ε kl + 2 µ  δ ik δ jl ε kl + δ ilδ jk ε kl  − β (θ − θ 0 ) δ ij
=ε 2 = ε= ε 
2 = ∆θ
ll ji ij
= ε ij
 The resulting constitutive equation is,
σ λ Tr ( ε ) 1 + 2 µ ε − β ∆θ 1
=  Isotropic linear thermoelastic
 
σ
=
 ij λδ ε
ij ll + 2 µ ε ij − β ∆θ δ ij i , j ∈ {1, 2,3} 
constitutive equation.

70
Inversion of the Constitutive Equation
1. ε is isolated from the Generalized Hooke’s Law for linear
thermoelastic problems:
=ε C −1 : σ + ∆θ C −1
 :β
σ= C : ε − β ∆θ
α
2. The thermal expansion coefficients tensor α is defined as:
def
α = C −1 : β

It is a 2nd order symmetric tensor which involves


6 thermal expansion coefficients

3. The inverse constitutive equation is obtained:


=ε C −1 : σ + ∆θ α

71
Inverse Isotropic Linear Thermoelastic
Constitutive Equation
 For the isotropic case:
 −1 ν 1 +ν
 C =− 1 ⊗ 1 + I
E E 1 − 2ν
 → α =
C −1
: ( β 1) = β1
ν 1 +ν
C −1 =
 ijkl

E
δ δ
ij kl +
E
( δ δ
ik jl + δ δ
il jk ) i , j , k .l ∈ {1, 2,3}
E

 The inverse const. eq. is re-written:

 ν 1 +ν
 ε = − Tr ( σ ) 1 + σ + α ∆θ 1
Inverse isotropic linear thermo
E E
 elastic constitutive equation.
ε = ν 1 +ν
− σ δ + σ ij + α ∆θ δ ij i, j ∈ {1, 2,3}
 ij E
ll ij
E

 Where α is a scalar thermal expansion coefficient related to the


scalar thermal constant parameter β through:
1 − 2ν
α = β
E
72
Thermal Stress
 Comparing the constitutive equations,

=σ λ Tr ( ε ) 1 + 2 µ ε Isotropic linear elastic constitutive equation.

=σ λ Tr ( ε ) 1 + 2 µ ε − β ∆θ 1 Isotropic linear thermoelastic constitutive equation.


t
= σ
nt = σ

the decomposition is made:


=
σ σ nt − σ t
Where:
 σ nt is the non-thermal stress: the stress produced if there is no
temperature increment.
 σ t is the thermal stress: the “corrector” stress due to the
temperature increment.

73
Thermal Strain
 Similarly, by comparing the inverse constitutive equations,
ν 1 +ν Inverse isotropic linear elastic
ε=− Tr ( σ ) 1 + σ
E E constitutive eq.
ν 1 +ν
ε=− Tr ( σ ) 1 + σ + α ∆θ 1 Inverse isotropic linear thermoelastic
E E nt
=ε = ε t
constitutive eq.
the decomposition is made:
=
ε ε nt + ε t
Where:
 ε nt is the non-thermal strain: the strain produced if there is no
temperature increment.
 ε t is the thermal strain: the “corrector” strain due to the
temperature increment.

74
Thermal Stress and Strain
 The thermal components appear when thermal processes are considered.

NON-THERMAL THERMAL
TOTAL
COMPONENT COMPONENT
σ nt = C : ε σ t = ∆θ β
=
σ σ nt − σ t Isotropic material: Isotropic material:
= σ nt λTr (ε) 1 + 2 µ ε σ t= β ∆θ 1

ε nt = C −1 : σ ε t = ∆θ α
=
ε ε nt + ε t Isotropic material:
ν 1 +ν Isotropic material:
− Tr (σ ) 1 +
ε =
nt
σ ε t= α ∆θ 1
E E
These are the equations used
in FEM codes. =σ C=: ε nt C :  ε − ε t 
= −1
ε C= : σ nt C −1 :  σ + σ t 

75
Thermal Stress and Strain

REMARK 1
In thermoelastic problems, a state of zero strain in a body does not necessarily
imply zero stress.
ε=0 → σ nt =0

σ =−σ t =− β∆θ 1 ≠ 0

REMARK 2
In thermoelastic problems, a state of zero stress in a body does not necessarily
imply zero strain.
σ =0 → ε nt =0

ε= α θ1≠0
ε t =∆

76
6.6 Thermal Analogies
Ch.6. Linear Elasticity

77
Solution to the Linear Thermoelastic
Problem
 To solve the isotropic linear thermoelastic problem posed thermal
analogies are used.
 The thermoelastic problem is solved like an elastic problem and then,
the results are “corrected” to account for the temperature effects.
 They use the same strategies and methodologies seen in solving
isotropic linear elastic problems:
 Displacement Formulation - Navier Equations.
 Stress Formulation - Beltrami-Michell Equations.
 Two basic analogies for solving quasi-static isotropic linear
thermoelastic problems are presented:
 1st thermal analogy – Duhamel-Neumann analogy.
 2nd thermal analogy

78
1st Thermal Analogy
 The governing eqns. of the quasi-static isotropic linear thermoelastic
problem are:
∇ ⋅ σ ( x, t ) + ρ0b ( x, t ) =
0 Equilibrium Equation

( x, t ) C : ε ( x, t ) − β ∆θ 1
σ= Constitutive Equation
1
ε ( x, t )= ∇ S u ( x, t )= ( u ⊗ ∇ + ∇ ⊗ u ) Geometric Equation
2

Γu : u =
u*
Boundary Conditions in Space
Γσ : t * =
σ ⋅n

79
1st Thermal Analogy
 The actions and responses of the problem are:
not
(I )
( x, t )
not
ACTIONS = A RESPONSES = R ( ) ( x, t )
I

b ( x, t )
u ( x, t )
t * ( x, t ) Elastic model ε ( x, t )
u * ( x, t ) EDPs+BCs σ ( x, t )
∆θ ( x, t )

REMARK
∆θ ( x,t ) is known a priori, i.e., it is
independent of the mechanical response.
This is an uncoupled thermoelastic problem.

80
1st Thermal Analogy
 To solve the problem following the methods used in linear elastic
problems, the thermal term must be removed.
 The stress tensor is split into =
σ σ nt − σ t and replaced into the governing
equations:
 Momentum equations
=σ σ nt − σ t ∇= ⋅ σ ∇ ⋅ σ nt − ∇ ⋅  σ=t
∇ ⋅ σ nt − ∇ ( β ∆θ )
β ∆θ 1

∇ ⋅ σ + ρ0 b =0
∇ ⋅ σ nt + ρ0 bˆ =
0
∇ ⋅ σ nt + ρ0  b − ∇ ( β ∆θ )  =
1
0
 ρ 0  1
b − ∇ ( β ∆θ )
bˆ =
not ρ0
= bˆ

81
1st Thermal Analogy
 Boundary equations:
σ σ nt − σ t
=
σ nt ⋅ n − σ t ⋅ n =t* σ nt ⋅ n = tˆ*
σ ⋅n =t *
Γσ :
σ nt ⋅ n = t* + σ t
⋅ n = t* + ( β∆θ ) n tˆ* = t* + ( β∆θ ) n
   
β∆θ 1⋅n tˆ*
ANALOGOUS PROBLEM – A linear elastic problem can be solved as:
1
∇ ⋅ σ nt + ρ0bˆ = b − ∇ ( β∆θ )
0 with bˆ = Equilibrium Equation
ρ0
=
σ nt
C= : ε λTr (ε) 1 + 2 µ ε Constitutive Equation
1
ε ( x, t )= ∇ S u ( x, t )= (u ⊗ ∇ + ∇ ⊗ u ) Geometric Equation
2

Γu : u = u*
Boundary Conditions in Space
Γσ : σ nt ⋅ n =tˆ* with tˆ* = t* + β∆θ n

82
1st Thermal Analogy
 The actions and responses of the ANALOGOUS NON-THERMAL PROBLEM are:
not
( II )
( x, t )
not
ACTIONS = A RESPONSES = R ( ) ( x, t )
II

bˆ ( x, t ) u ( x, t )
Elastic model ε ( x, t )
tˆ* ( x, t )
EDPs+BCs σ nt ( x, t )
u* ( x, t )

ORIGINAL PROBLEM (I) ANALOGOUS (ELASTIC)


PROBLEM (II)
83
1st Thermal Analogy
 If the actions and responses of the original and analogous problems are
compared:
≡ b 1 
ACTIONS  b   bˆ   b − bˆ  ρ ∇ ( β ∆θ ) 
 u∗   ∗   
  def ( III )
0
  u   0 
A ( ) ( x, t ) − A ( ) ( x, t ) =  *  −  *  =  * ˆ*  = = A
I II
 0
 t   tˆ  t − t ≡ t *  − ( β∆θ ) n 
∆ θ   0   ∆θ   
     ∆θ 

RESPONSES
u   u   0   0  def
       
R ( ) (x, t ) − R ( ) (x, t ) = ε  −  ε  = 0  =  0  =R ( )
I II III

σ  σ nt  σ − σ nt  − β∆θ 1
     t  
= −σ

Responses are proven to be the solution of a thermoelastic


problem under actions

84
1st Thermal Analogy

THERMOELASTIC ANALOGOUS THERMOELASTIC


ORIGINAL ELASTIC (TRIVIAL)
PROBLEM (I) PROBLEM (II) PROBLEM (III)
1
b ( x, t ) b − ∇ ( β∆θ )
bˆ ( x, t ) = =b
1
∇ ( β∆θ )
ρ0 ρ0
t* ( x, t )
A ( I ) ( x, t ) u ( x, t )
* A ( II ) ( x, t ) tˆ* ( x, t ) = t* + ( β∆θ ) n A ( III ) ( x, t ) t* = − ( β∆θ ) n
u* ( x, t ) u ∗ = 0
∆θ ( x, t )
∆θ =0 ∆ θ ( x,t )

u ( x, t ) u ( x, t ) u=0
R ( I ) ( x, t ) ε ( x, t ) R(
II )
( x, t ) ε ( x, t )
R(
III )
( x, t ) ε=0
σ ( x, t ) σ nt ( x, t ) σ=−σ t =− ( β∆θ ) 1

86
2nd Thermal Analogy
 The governing equations of the quasi-static isotropic linear thermoelastic
problem are:
∇ ⋅ σ ( x, t ) + ρ0b ( x, t ) =
0 Equilibrium Equation

ε ( x, t ) C -1 : σ ( x, t ) + α ∆θ 1
= Inverse Constitutive Equation
1
ε ( x, t )= ∇ S u ( x, t )= ( u ⊗ ∇ + ∇ ⊗ u ) Geometric Equation
2

Γu : u =
u*
Boundary Conditions in Space
Γσ : t * =
σ ⋅n

87
2nd Thermal Analogy
 The actions and responses of the problem are:
not
(I )
( x, t )
not
ACTIONS = A RESPONSES = R ( ) ( x, t )
I

b ( x, t )
u ( x, t )
t * ( x, t ) Elastic model ε ( x, t )
u * ( x, t ) EDPs+BCs σ ( x, t )
∆θ ( x, t )

REMARK
∆θ ( x,t ) is known a priori, i.e., it is
independent of the mechanical response.
This is an uncoupled thermoelastic problem.

88
2nd Thermal Analogy
The assumption is made that ∆θ (x, t ) and α (x) are such that the thermal
strain field ε t = α∆θ 1 is integrable (satisfies the compatibility equations).
 If the thermal strain field is integrable, there exists a field of thermal
displacements, ut ( x, t ), which satisfies:
1 t
ε t ( x, t ) = (α ∆θ ) 1 = ∇ S ut =( u ⊗ ∇ + ∇ ⊗ ut )
2
1  ∂ uit ∂ u tj 
(α ∆θ ) δ ij =
ε ijt = +  i, j ∈ {1, 2,3}
2  ∂ x j ∂ xi 
REMARK
The solution u t ( x, t ) is determined except for a rigid body motion
characterized by a rotation tensor Ω ∗ and a displacement vector c* .
The family of admissible solutions is u t (=x, t ) u ( x, t ) + Ω ∗ ⋅ x + c* .
This movement can be arbitrarily chosen (at convenience).
 Then, the total displacement field is decomposed by defining:
def
u =
( x , t ) u( x , t ) − u t ( x , t )
nt
=
u u nt + ut
89
2nd Thermal Analogy
 To solve the problem following the methods used in linear elastic
problems, the thermal terms must be removed.
 The strain tensor and the displacement vector splits,=
ε ε nt + ε t and =
u u nt + ut
are replaced into the governing equations:
 Geometric equations:
ε= ∇ S u= ∇ S ( u nt + ut =
) ∇ S u nt + ∇
 u= ∇
S t
 u + εt
S nt

εt εnt ε nt = ∇ S u nt
=
ε ε nt + ε t

 Boundary equations:

u = u*
u nt + ut =
u* Γu : u =u* u nt =
u* − u t
=
u u nt + ut

90
2nd Thermal Analogy

ANALOGOUS PROBLEM – A linear elastic problem can be solved as:

∇ ⋅ σ + ρ 0b =0 Equilibrium Equation
ε nt = C -1 : σ Inverse constitutive Equation
ε nt = ∇ S u nt Geometric Equation

Γu : u nt =
u* − u t
Boundary Conditions in space
Γσ : σ ⋅ n =t*

91
2nd Thermal Analogy
 The actions and responses of the ANALOGOUS PROBLEM are:
not
( II )
( x, t )
not
ACTIONS = A RESPONSES = R ( ) ( x, t )
II

b ( x, t ) u nt ( x, t )
Elastic model ε nt ( x, t )
t* ( x, t ) EDPs+BCs σ ( x, t )
u nt u* ( x, t ) − ut ( x, t )
=

ORIGINAL PROBLEM (I) ANALOGOUS PROBLEM (II)

92
2nd Thermal Analogy
 If the actions and responses of the original and analogous problems are
compared:
b   b   0 
ACTIONS  u∗  u∗ − ut   ut  def
(I ) ( II )      
A (x, t ) − A (x, t ) =
 *   * =
−  = A ( III )
t   t   0 
∆ θ   0  ∆ θ 

RESPONSES
 u  u  u   ut 
nt t

(I ) ( II )    nt   t    def ( III )
R (x, t ) − R (x, t ) =  ε  −  ε  =  ε  = α ∆θ 1 = R
σ   σ   0   0 
       

Responses are proven to be the solution of a thermo-elastic


problem under actions

93
2nd Thermal Analogy

THERMOELASTIC ANALOGOUS THERMOELASTIC


ORIGINAL ELASTIC TRIVIAL
PROBLEM (I) PROBLEM (II) PROBLEM (III)

b ( x, t ) b b = 0
t* ( x, t ) = u∗ − u t
u u ∗ = ut ( x, t )
A ( I ) ( x, t ) u ( x, t )
* A(
II )
( x, t ) t*
A ( III ) ( x, t ) t* = 0
∆θ ( x, t ) ∆θ =0 ∆ θ ( x, t )

u ( x, t ) u nt ( x, t ) u = u t ( x, t )
R ( I ) ( x, t ) ε ( x, t ) R ( II ) ( x, t ) ε nt ( x, t )
R(
III )
( x, t ) ε= ε=t
(α ∆θ ) 1
σ ( x, t ) σ ( x, t ) σ=0

95
2nd Analogy in structural analysis

u x = u xt = α∆θ x u x = u xnt

ε x = ε xt = α∆θ ε x = ε xnt
Γu : ux =
u xt α∆θ 
= Γu : u x =u*x − u xt =−α∆θ 
x =
x =
0

96
Thermal Analogies
 Although the 2nd analogy is more commonly used , the 1st analogy
requires less corrections.
 The 2nd analogy can only be applied if the thermal strain field
is integrable.
 It is also recommended that the integration be simple.
 The particular case
 Homogeneous material:= α (x) const
=. α
 Lineal thermal increment: ∆θ = ax + by + cz + d
is of special interest because the thermal strains are:
ε t = α ∆θ 1 = linear polinomial
and trivially satisfy the compatibility conditions (involving second
order derivatives).
97
Thermal Analogies
 In the particular case
 Homogeneous material: =α (x) const
=. α
 Constant thermal increment: ∆θ (x) = ∆θ
const. =
the integration of the strain field has a trivial solution because
the thermal strains are constant ε t = ∆θ α 1 = const. , therefore:
rigid body motion
(can be chosen arbitrarily:
u ( x, t ) = α ∆θ x + Ω ⋅ x + c
t ∗ ∗
at convenience)
The thermal displacement is:
ut ( x, t )= α ∆θ x x + ut = x + α ∆θ x = (1 + α ∆θ ) x
HOMOTHECY
(free thermal expansion)

98
6.7 Superposition Principle
Ch.6. Linear Elasticity

99
Linear Thermoelastic Problem
 The governing eqns. of the isotropic linear thermoelastic problem are:

∇ ⋅ σ ( x, t ) + ρ0b ( x, t ) =
0 Equilibrium Equation

( x, t ) C : ε ( x, t ) − β ∆θ 1
σ= Constitutive Equation
1
ε ( x, t )= ∇ S u ( x, t )= ( u ⊗ ∇ + ∇ ⊗ u ) Geometric Equation
2

Γu : u =
u*
Boundary Conditions in space
Γσ : t * =
σ ⋅n

u ( x, 0 ) = 0
Initial Conditions
u ( x, 0 ) = v 0

100
Linear Thermoelastic Problem
 Consider two possible systems of actions:
b ( ) ( x, t ) b( ( x, t )
2)
1

t *( ) ( x, t ) t*( ) ( x, t )
1 2

A (1) ( x,t ) ≡ u*( ) ( x, t )


1 A(
2)
( x,t ) ≡ u*( ) ( x, t )
2

∆θ (1) ( x, t ) ∆θ ( 2) ( x, t )
v 0( ) ( x ) v 0( ) ( x )
1 2

and their responses :


u(1) ( x, t ) u( ( x, t )
2)

R ( ) ( x,t ) ≡
1
ε ( ) ( x, t )
1
R(
2)
( x,t ) ≡ ε ( ) ( x, t )
2

σ ( ) ( x, t ) σ ( ) ( x, t )
1 2

101
Superposition Principle
 The solution to the system of actions= A (3) λ (1) A (1) + λ ( 2) A ( 2) where
λ (1)and λ ( 2) are two given scalar values, is
= R (3) λ (1) R (1) + λ ( 2) R ( 2) .

The response to the lineal thermoelastic problem caused by two or more


groups of actions is the lineal combination of the responses caused by each
action individually.

 This can be proven by simple substitution of the linear


combination of actions and responses into the governing
equations and boundary conditions.

 When dealing with non-linear problems (plasticity, finite


deformations, etc), this principle is no longer valid.

102
6.8 Hooke’s Law in Voigt Notation
Ch.6. Linear Elasticity

103
Stress and Strain Vectors
 Taking into account the symmetry of the stress and strain tensors,
these can be written in vector form: εx 
 1 1  ε 
 εx 2
γ xy γ xz
2 
 y
 ε x ε xy ε xz    def  ε 

=

ε xy ε y ε yz 
ε =
 not .  1
γ εy
1 
γ yz {ε} γ z  ∈ R 6
=
 2 xy 2   xy 
ε xz ε yz ε z   
  γ xz 
 1 γ xz 1
γ yz εz   
 2 2  γ yz 
REMARK VOIGT
The double contraction ( σ : ε ) is NOTATION σ x 
σ 
transformed into the scalar (dot)  y
σ x τ xy τ xz 
product ({σ} ⋅ {ε} ) :   def σ 

vectors
σ ≡ τ xy σ y τ yz  {σ} τ z  ∈ R 6
=
τ xz τ yz σ z   xy 
σ=
:ε {σ} ⋅ {ε} σ ij ε ij = σ iε i  τ xz 
2nd order  
tensors τ yz 

104
Inverse Constitutive Equation
 The inverse constitutive equation is rewritten:
ν 1 +ν
ε=− Tr ( σ ) 1 + σ + α ∆θ 1 {ε} = Cˆ −1 ⋅ {σ} + {ε}
t

E E

Where Cˆ −1 is an elastic constants inverse matrix and {ε} is a thermal


t

strain vector: α∆θ 0 0 


 1 −ν −ν
0 0 0
 ε t ≡  0 α∆θ 0 
E

E E
  0 0 α∆θ 
 −ν 1 −ν
0 0 0
E E E 
 −ν −ν 1 
 0 0 0
ˆ −1 =  E
C
E E 
α∆θ 
 
 0 α∆θ 
0 
1
 0 0
G
0  
  α∆θ 
{ε} =  
t
 0 1
0 0 0 0
 G   0 
 1  0 
 0 0 0 0 0   
 G
 0 

105
Hooke’s Law
 By inverting the inverse constitutive equation, Hooke’s Law in terms of the
stress and strain vectors is obtained:

{σ} =Cˆ ⋅ ({ε} − {ε} )


t

 ν ν 
 1 1 −ν 1 −ν
0 0 0 
Where Ĉ is an elastic  
 ν 1
ν
0 0 0 
constants matrix : 1 −ν 1 −ν 
 
 ν ν
1 0 0 0 
E (1 −ν ) 1 −ν 1 −ν 
C=
ˆ  1 − 2ν 
(1 +ν )(1 − 2ν )  0 0 0 0 0 
 2 (1 −ν ) 
 1 − 2ν 
 0 0 0 0 0 
 2 (1 −ν ) 
 
 0 1 − 2ν 
0 0 0 0
 2 (1 −ν ) 

106
Chapter 6
Linear Elasticity

rs
ee
s gin
6.1 Hypothesis of the Linear Theory of Elasticity

t d le En

r
The linear theory of elasticity can be considered a simplification of the general

ba
ge ro or
eS m
theory of elasticity, but a close enough approximation for most engineering ap-

ci
f
plications. The simplifying hypotheses of the linear theory of elasticity are

ra
C d P cs
b
a
a) Infinitesimal strains. The displacements and its gradients are small, see
i
an an n

Chapter 2.
y ha

• Small displacements. The material configuration (corresponding to the


le
liv or ec

reference time t0 ) is indistinguishable from the spatial one (correspond-


ing to the present time t) and, consequently, the spatial and material co-
M

.A

ordinates cannot be distinguished from each other either, see Figure 6.1.
m

d
uu

x = X+ u =⇒ x≈X
e

(6.1)

X Th

er
tin

≈0
on

.O
C

Figure 6.1: Small displacements are considered in the linear theory of elasticity.

263
264 C HAPTER 6. L INEAR E LASTICITY

From (6.1), one can write


∂x
F= =1 =⇒ |F| ≈ 1 . (6.2)
∂X

Remark 6.1. As a consequence of (6.1), there is no difference be-


tween the spatial and material descriptions of a property,

x=X =⇒ γ (x,t) = γ (X,t) = Γ (X,t) = Γ (x,t) ,

rs
and all references to the spatial and material descriptions (in addition

ee
to any associated concepts such as local derivative, material deriva-

s gin
tive, etc.) no longer make sense in infinitesimal elasticity.
Likewise, the spatial Nabla differential
  operator ( ∇ ) is indistin-

t d le En
guishable from the material one ∇ ¯ ,

r
ba
ge ro or
∂ (•) ∂ (•)

eS m
= =⇒ ¯ (•) .
∇ (•) = ∇

ci
∂X ∂x f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec

Remark 6.2. As a consequence of (6.2) and the principle of conser-


M

.A

vation of mass, the density in the present configuration ρt ≡ ρ (X,t)


coincides with the one in the reference configuration ρ0 ≡ ρ (X, 0)
m

(which is assumed to be known),


d
uu
e

ρ0 = ρt |F| ≈ ρt ,
X Th

er
tin

and, therefore, the density is not an unknown in linear elasticity


on

.O

problems.
C

• Small displacement gradients. As a consequence, no distinction is made


between the material strain tensor E (X,t) and the spatial strain tensor
e (x,t), which collapse into the infinitesimal strain tensor ε (x,t).

E (X,t) ≈ e (x,t) = ε (x,t)



⎪ 1

⎨ ε = ∇ u = (u ⊗ ∇ + ∇ ⊗ u)
S
(6.3)
2

⎪ 1 ∂ ui ∂ u j
⎩ εi j = + i, j ∈ {1, 2, 3}
2 ∂ x j ∂ xi

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Linear Elastic Constitutive Equation. Generalized Hooke’s Law 265

b) Existence of a neutral state. The existence of a neutral state in which the


strains and stresses are null is accepted. Usually, the neutral state is under-
stood to occur in the reference configuration.

ε (x,t0 ) = 0
(6.4)
σ (x,t0 ) = 0

c) The deformation process is considered (in principle) to be isothermal1


and adiabatic.

rs
ee
Definition 6.1. Isothermal processes are those that take place at a
temperature θ (x,t) that is constant along time,

s gin
θ (x,t) ≡ θ (x) .

t d le En

r
Adiabatic processes are those that take place without heat generation

ba
ge ro or
eS m
at any point and instant of time.

ci
f

ra
C d P cs
b
a
Heat generated inside a domain V per unit of time:
 
i
an an n

Qe = ρr dV − q · n dS = 0 ∀ΔV ⊂ V
y ha

le
V ∂V
liv or ec

=⇒ ρr − ∇ · q = 0 ∀x ∀t
M

.A

Slow deformation processes are commonly considered to be adia-


m

batic.
d
uu
e
X Th

er
tin
on

.O

6.2 Linear Elastic Constitutive Equation. Generalized


C

Hooke’s Law
©

Hooke’s law for unidimensional problems establishes the proportionality be-


tween the stress, σ , and the strain, ε, by means of the constant named elastic
modulus, E,
σ = Eε . (6.5)
In the theory of elasticity, this proportionality is generalized to the multidimen-
sional case by assuming the linearity of the relation between the components of
the stress tensor σ and those of the strain tensor ε in the expression known as
generalized Hooke’s law,
1 The restriction to isothermal processes disappears in the linear theory of thermoelasticity,
which will be addressed in Section 6.6.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
266 C HAPTER 6. L INEAR E LASTICITY


⎨ σ (x,t) = C : ε (x,t)
Generalized (6.6)
Hooke’s law ⎩ σ = C ε i, j ∈ {1, 2, 3}
ij i jkl kl

which constitutes the constitutive equation of a linear elastic material.


The fourth-order tensor C (denoted as tensor of elastic constants) has 34 = 81
components. However, due to the symmetry of the tensors σ and ε , it must
exhibit certain symmetries in relation to the exchange of its indexes. These are:

Ci jkl = C jikl

rs
→ minor symmetries
Ci jkl = Ci jlk

ee
(6.7)

s gin
Ci jkl = Ckli j → major symmetries

t d le En
Consequently, the number of different constants in the tensor of elastic constants
C is reduced to 21.

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
Remark 6.3. An essential characteristic of the elastic behavior
b
a
i
(which is verified in (6.5)) is that the stresses at a certain point and
an an n

time, σ (x,t), depend (only) on the strains at said point and time,
y ha

ε (x,t), and not on the history of previous strains.


le
liv or ec
M

.A
m

6.2.1 Elastic Potential


d
uu
e

Consider the specific internal energy u (x,t) (internal energy per unit of mass)
X Th

er
tin

and the density of internal energy û (x,t) (internal energy per unit of volume),
which related through
on

.O

û (x,t) = ρ0 u (x,t) ,
C


  
©

(6.8)
du du d (ρ0 u) d û
ρ ≈ ρ0 = = ,
dt dt dt dt
where ρ0 ≈ ρ (see Remark 6.2) has been taken into account. Consider now the
energy equation in its local form2 ,
du d û . d û .
ρ0 = = σ : d + ρ0 r − ∇ · q = σ : ε =⇒ =σ :ε , (6.9)
dt dt dt

.
2 The identity d = ε , characteristic of the infinitesimal strain case, is considered here.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Linear Elastic Constitutive Equation. Generalized Hooke’s Law 267

where the adiabatic nature of the deformation process (ρ0 r − ∇ · q = 0) has been
considered. Then, the global (integral) form of the energy equation in (6.9) is
obtained by integrating over the material volume V .

Global form of the energy equation in linear elasticity


  
dU d d û .
= û dV = dV = σ : ε dV
dt dt dt (6.10)
Vt ≡V V V

with U (t) = û (x,t) dV

rs
ee
V

s gin
Here, U (t) is the internal energy of the material volume considered.

t d le En

r
ba
ge ro or
Remark 6.4. The stress power (in the case of linear elasticity) is an

eS m
ci
exact differential,
f

ra
C d P cs
 . dU
b
a
stress power = σ : ε dV = .
i
dt
an an n

V
y ha

le
liv or ec

Replacing now (6.6) in (6.9),


M

.A

i↔k
j↔l
m

1. .   
d

d û not . . . .
uu

= û = σ : ε = εi j σi j = εi j Ci jkl εkl = εi j Ci jkl εkl + εi j Ci jkl εkl =


e

dt 2
X Th

er

1.  1 .
tin

. . 
= εi j Ci jkl εkl + εkl Ckli j εi j = εi j Ci jkl εkl + εi j Ci jkl εkl =
on

2 2
.O

1d   1d
C

= εi j Ci jkl εkl = (εε : C : ε ) ,


2 dt 2 dt
©

(6.11)
where the symmetries in (6.7) have been taken into account. Integrating the ex-
pression obtained and imposing the condition that the density of internal energy
û (x,t0 ) in the neutral state be null3 (for t = t0 ⇒ ε (x,t0 ) = 0) produces the
density of internal energy.

3 The condition û (x,t0 ) = 0 can be introduced without loss of generality.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
268 C HAPTER 6. L INEAR E LASTICITY


1 ⎪
û (x,t) = (εε (x,t) : C : ε (x,t)) + a (x) ⎬
2 =⇒


û (x,t0 ) = 0 ∀x
(6.12)
1
=⇒ ε (x,t0 ) : C : ε (x,t0 ) + a (x) = a (x) = 0 ∀x
2   
=0
 1 1
Density of
internal energy û (εε ) = (εε : C : ε ) = εi j Ci jkl εkl (6.13)
2 2

rs
ee
Now, (6.13) is differentiated with respect to ε , considering once more the

s gin
symmetries in (6.7).

t d le En

⎪ ∂ û (εε ) 1

⎪ 1 1 1
⎨ = C :ε + ε :C = C :ε + C :ε =C :ε =σ

r
∂ε

ba
2 2 2 2

ge ro or
eS m

ci

⎪ ∂ û (εε ) 1
⎪ 1 1
f 1

ra
⎩ = Ci jkl εkl + εkl Ckli j = Ci jkl εkl + Ci jkl εkl = Ci jkl εkl = σi j
C d P cs
∂ εi j
b
2 2 2 2

a
i
⎧ (6.14)
an an n


y ha


⎪ ε
∂ û (ε )
⎨ =σ
le
∂ε
liv or ec

=⇒ (6.15)

⎪ ∂ û ε
(ε )
M


.A

⎩ = σi j i, j ∈ {1, 2, 3}
∂ εi j
m

Equation (6.15) qualifies the density of internal energy û (εε ) as a potential


uu
e

for the stresses (which are obtained by differentiation of this potential), named
X Th

er
tin

elastic potential.

on

.O


⎪ 1 1

⎨ û (εε ) = 2 ε : C
⎪ :ε = σ :ε
C

   2
©

Elastic potential =σ (6.16)




⎪ ε
⎩ ∂ û (ε ) = σ

∂ε

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Isotropy. Lamé’s Constants. Hooke’s Law for Isotropic Linear Elasticity 269

6.3 Isotropy. Lamé’s Constants. Hooke’s Law for Isotropic


Linear Elasticity

Definition 6.2. An isotropic material is that which has the same


properties in all directions.

rs
The elastic properties of a linear elastic material are contained in the tensor of
elastic constants C defined in (6.6) and (6.7). Consequently, the components of

ee
this tensor must be independent of the orientation of the Cartesian system used4 .

s gin
Consider, for example, the systems {x1 , x2 , x3 } and {x1 , x2 , x3 } in Figure 6.2,
the constitutive equation for these two systems is written as

t d le En

r
ba
σ ] = [C
{x1 , x2 , x3 } =⇒ [σ C] : [εε ]

ge ro or
eS m
ci
(6.17)
f

ra
C d P cs
{x1 , x2 , x3 } =⇒ σ]
[σ C]
= [C : [εε ]
b
a
i
an an n
y ha

and, for the case of an isotropic material, the components of C in both sys-
tems must be the same ( [C C] = [C C] ). Therefore, the aforementioned definition
le
liv or ec

of isotropy, which has a physical character, translates into the isotropic charac-
M

.A

ter, in the mathematical sense, of the tensor of elastic constants C .



m


d


⎨ C = λ 1 ⊗ 1 + 2μI

uu
e

Tensor of elastic  
X Th

er

Ci jkl = λ δi j δkl + μ δik δ jl + δil δ jk (6.18)


tin

constants ⎪


⎩ i, j, k, l ∈ {1, 2, 3}
on

.O
C

Here, λ and μ are known as Lamé’s constants, which characterize the elastic
©

behavior of the material and must be obtained experimentally.

Remark 6.5. The isotropy condition reduces the number of elastic


constants of the material from 21 to 2.

4 A tensor is isotropic if it maintains its components in any Cartesian coordinate system. The
most general expression of a fourth-order isotropic tensor is C = λ 1⊗1+2μI , ∀λ , μ. Here,
the fourth-order
 symmetric (isotropic) unit tensor I is defined by means of its components as
[I]i jkl = δik δ jl + δil δ jk /2.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
270 C HAPTER 6. L INEAR E LASTICITY

Figure 6.2: Representation of the Cartesian systems {x1 , x2 , x3 } and {x1 , x2 , x3 }.

rs
ee
Introducing (6.18) in (6.6) results in the isotropic linear elastic constitutive

s gin
equation,
1 

t d le En
1
σi j = Ci jkl εkl = λ δi j δkl εkl +2μ δik δ jl εkl + δil δ jk εkl . (6.19)
   2    2   

r
ba
ge ro or
εll εi j ε ji = εi j

eS m
  

ci
f εi j

ra
C d P cs
b
a

i
an an n

Constitutive eqn. for a ⎨ σ = λ Tr (εε ) 1 + 2μεε


y ha

linear elastic material. (6.20)


⎩ σi j = λ δi j εll + 2μεi j
le
i, j ∈ {1, 2, 3}
liv or ec

Hooke’s law.
M

.A
m

6.3.1 Inversion of Hooke’s Law. Young’s Modulus. Poisson’s Ratio


d
uu
e

The constitutive equation (6.20) provides the stresses in terms of the strains. To
X Th

er
tin

obtain its inverse expression, the following procedure is followed.


on

.O

a) The trace of (6.20) is obtained,



C

σ ) = λ Tr (εε ) Tr (1) +2μ Tr (εε ) = (3λ + 2μ) Tr (εε ) ⎪


Tr (σ ⎪

©

   ⎪

3 =⇒
(i = j) =⇒ σii = λ εll δii +2μεii = (3λ + 2μ) εll ⎪


⎪ (6.21)
 ⎭
3
1
=⇒ Tr (εε ) = σ) .
Tr (σ
(3λ + 2μ)
b) ε is isolated from (6.20) and introduced in (6.21),
1 1 λ 1
ε =− λ Tr (εε ) 1 + σ =− σ)1+
Tr (σ σ. (6.22)
2μ 2μ 2μ (3λ + 2μ) 2μ

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Isotropy. Lamé’s Constants. Hooke’s Law for Isotropic Linear Elasticity 271

The new elastic properties E (Young’s modulus) and ν (Poisson’s ratio) are de-
fined as follows.

Young’s modulus or μ (3λ + 2μ) ⎪⎪


E= ⎬
tensile (elastic) modulus λ +μ =⇒


λ ⎪

Poisson’s ratio ν= ⎭
2 (λ + μ)
⎧ (6.23)

rs

⎪ νE
⎨λ =

ee
(1 + ν) (1 − 2ν)
=⇒

s gin

⎪ E

⎩μ = =G shear (elastic) modulus
2 (1 + ν)

t d le En

r
Equation (6.22) can be expressed in terms of E and ν, resulting in the inverse

ba
ge ro or
eS m
Hooke’s law.

ci
f

ra

C d P cs
b
a

⎪ ν 1+ν

i
⎪ ε = − Tr (σ σ)1+ σ
an an n

Inverse constitutive ⎨⎪ E E
y ha

equation for an isotropic ν 1+ν (6.24)


linear elastic material ⎪
le
⎪ εi j = − σll δi j + σi j
liv or ec


⎪ E E

⎩ i, j ∈ {1, 2, 3}
M

.A
m

d
uu

Finally, (6.24) is rewritten, using engineering notation for the components of the
e

strain and stress tensors.


X Th

er
tin

1 1
on

.O

εx = (σx − ν (σy + σz )) γxy = τxy


E G
C

1 1
©

εy = (σy − ν (σx + σz )) γxz = τxz (6.25)


E G
1 1
εz = (σz − ν (σx + σy )) γyz = τyz
E G

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
272 C HAPTER 6. L INEAR E LASTICITY

Example 6.1 – Consider an uniaxial tensile test of a rectangular cuboid


composed of an isotropic linear elastic material with Young’s modulus E and
shear modulus G, such that its uniform stress state results in

σx = 0 and σy = σz = τxy = τxz = τyz = 0 .


Obtain the strains in engineering notation.

Solution
From (6.25) one obtains

rs
⎧ ⎧
⎪ σx ⎪ τxy
⎪ ⎪ γxy = =0

ee
⎪ εx = ⎪

⎨ E ⎪
⎨ G
σ τ

s gin
σy = σz = 0 =⇒ εy = −ν x τxy = τxz = τyz = 0 =⇒ γxz = xz = 0

⎪ E ⎪
⎪ G
⎪ ⎪
⎩ εz = −ν σx
⎪ ⎩ γyz = τyz = 0

t d le En
E G

r
ba
ge ro or
eS m
Therefore, due to these strains, the rectangular cuboid subjected to an uni-

ci
f

ra
axial tensile test, shown in the figure below, stretches in the x-direction and
C d P cs
b
a
contracts in the y- and z-directions.
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O

6.4 Hooke’s Law in Spherical and Deviatoric Components


C

Consider the decomposition of the stress tensor σ and the deformation tensor ε
©

in their spherical and deviatoric parts,


1
σ= σ ) 1 + σ = σm 1 + σ ,
Tr (σ (6.26)
3
1 1
ε= Tr (εε ) 1 + ε = e1 + ε . (6.27)
3 3

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Hooke’s Law in Spherical and Deviatoric Components 273

The volumetric strain e = Tr (εε ) is obtained by computing the trace of (6.24).

ν 1+ν 1 − 2ν
e = Tr (εε ) = − σ ) Tr (1) +
Tr (σ σ) =
Tr (σ σ) =
Tr (σ
E    E E   
3 3σm (6.28)
3 (1 − 2ν)
= σm
E


⎪ E

⎨ σm = e = Ke

rs
=⇒ 3 (1 − 2ν) (6.29)

ee

⎪ de f 2 E
⎩K = λ + μ = = bulk modulus

s gin
3 3 (1 − 2ν)
Introducing (6.26), (6.27) and (6.29) in (6.24), results in

t d le En

r
ν 1+ν 1 − 2ν 1+ν

ba
ge ro or
ε = − 3σm 1 + (σm 1 + σ ) = σm 1 + σ =

eS m
E 

ci
E E E
f

ra
E
C d P cs
b
e

a
3 (1 − 2ν)
i
an an n

1+ν 1+ν (6.30)


y ha

1 1 1
= e1+ σ =⇒ ε = e 1 + ε = e 1 + σ
le
3 E 3 3 E
liv or ec
M

.A

1+ν 1 1
=⇒ ε = σ = σ = σ .
m

E 2μ 2G
d
uu
e

Equations (6.29) and (6.30) relate the spherical part (characterized by the mean
X Th

er
tin

σ and ε ) of the
stress σm and the volumetric strain e) and the deviatoric part (σ
stress and strain tensors as follows.
on

.O
C

σm = Ke Spherical part
©

 (6.31)
σ = 2Gεε
Deviatoric part
σ i j = 2Gε i j i, j ∈ {1, 2, 3}

Remark 6.6. Note the proportionality between σm and e as well as


between σ i j and ε i j (component to component), see Figure 6.3.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
274 C HAPTER 6. L INEAR E LASTICITY

Figure 6.3: Hooke’s law in spherical and deviatoric components.

rs
6.5 Limits in the Values of the Elastic Properties

ee
s gin
Thermodynamic considerations allow proving that the tensor of elastic constants
C is positive-definite5 , and, thus,

t d le En
ε :C :ε >0 ∀εε = 0 . (6.32)

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
Remark 6.7. As a consequence of (6.32), the elastic potential is al-
b
a
i
ways null or positive,
an an n

1
y ha

û (εε ) = ε : C : ε ≥ 0 .
2
le
liv or ec
M

.A
m

d
uu

Remark 6.8. The elastic potential has a minimum at the neutral state,
e

that is, for ε = 0 (see Figure 6.4). In effect, from (6.15),


X Th

er
tin

1 ∂ û (εε ) ∂ 2 û (εε )
û (εε ) = ε : C : ε , σ = = C : ε and =C .
on

.O

2 ∂ε ∂ε ⊗∂ε
Then, for ε = 0,
C


∂ û (εε ) 
©

 =0 =⇒ û (εε ) has an extreme


∂ ε ε =0 (maximum-minimum) at ε = 0.

∂ 2 û (εε ) 
= C =⇒ The extreme is a minimum.
∂ε ⊗∂ε  ε =0 
positive-
definite

5 A fourth-order symmetric tensor A is defined positive-definite if for all second-order


tensor x = 0 the expression x : A : x = xi j Ai jkl xkl > 0 is satisfied and, in addition,
x : A : x = 0 ⇔ x = 0.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Limits in the Values of the Elastic Properties 275

Figure 6.4: Elastic potential.

rs
ee
s gin
Consider the expression of the elastic potential (6.16) and the constitutive
equation (6.20), then,

t d le En
1 

r
1 1
û (εε ) = ε : C : ε = σ : ε = λ Tr (εε ) 1 + 2μεε : ε =

ba
ge ro or
eS m
2 2 2

ci
f (6.33)

ra
1 1
= λ Tr (εε ) 1 : ε +μεε : ε = λ Tr2 (εε ) + μεε : ε .
C d P cs
b
a
2  2
i
an an n

Tr (εε )
y ha

le
Expression (6.33) can also be written in terms of the spherical and deviatoric
liv or ec

components of strain6 ,
M

.A

1  2
m

1
û (εε ) = λ Tr (εε ) + μεε : ε = λ e2 + μεε : ε .
d

(6.34)
2   
uu

2
e

e
X Th

er
tin

Here, the double contraction of the infinitesimal strain tensor is


on

.O


1 1 1 2
e 1 + ε = e2 1 : 1 + e 1 : ε + ε : ε =
C

ε :ε = e1+ε :
 3   
©

3 3 9
3 Tr (εε ) = 0 (6.35)
1
= e2 + ε : ε .
3
Replacing (6.35) in (6.34),

1 2 1 2 1 2
û (εε ) = λ e + μe + μεε : ε = λ + μ e2 + μεε : ε . (6.36)
2 3 2 3
  
K
6 The trace of a deviatoric tensor is always null, Tr (εε ) = 0.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
276 C HAPTER 6. L INEAR E LASTICITY

1
û (εε ) = Ke2 + μεε : ε ≥ 0 (6.37)
2

Consider now an isotropic linear elastic material, characterized by a certain


value of its elastic properties. Equation (6.37) must be satisfied for any defor-
mation process. Consider two particular types:
a) A pure spherical deformation process

1 ⎪
ε = e 1⎬
(1)
1

rs
3 =⇒ û(1) = Ke2 ≥ 0 =⇒ K > 0 (6.38)
(1) ⎪ 2

ee
ε = 0 ⎭

s gin
b) A pure deviatoric deformation process7

t d le En

ε (2) = ε

r
=⇒ û(2) = μεε : ε ≥ 0 =⇒ μ > 0

ba
(6.39)

ge ro or
(2)

eS m
e =0

ci
f

ra
C d P cs
Equations (6.38) and (6.39) lead to
b
a
i
an an n

E E
K= >0 and μ =G= >0 (6.40)
y ha

3 (1 − 2ν) 2 (1 + ν)
le
liv or ec

which are the limits in the values of the elastic constants K and G. Experience
M

.A

proves that the Poisson’s ratio ν is always non-negative8 and, therefore


m


d
uu

E
> 0⎪

e

2 (1 + ν)
X Th

=⇒ E >0,
er
tin



ν ≥0
on

.O

⎫ (6.41)
E
> 0⎪

C

3 (1 − 2ν) 1
=⇒ 0≤ν ≤ .
©


⎭ 2
E ≥0

7 The double contraction or double dot product of a tensor by itself is always equal or greater
than zero: ε : ε = ε i j ε i j ≥ 0.
8 In rare cases, a material can have a negative Poisson’s ratio. Such materials are named
auxetic materials.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
The Linear Elastic Problem 277

Initial actions:

b (x, 0)
t =0 →
t (x, 0)

Actions along time t:



b (x,t)
t (x,t)

rs
Figure 6.5: Linear elastic problem.

ee
s gin
6.6 The Linear Elastic Problem

t d le En
Consider the linear elastic solid9 in Figure 6.5, which is subjected to certain ac-

r
ba
ge ro or
tions characterized by the vector of body forces b (x,t) in the interior of the vol-

eS m
ci
ume V and the traction vector t (x,t) on the boundary ∂V . The set of equations
f

ra
that allow determining the evolution along time of the displacements u (x,t),
C d P cs
b
a
strains ε (x,t) and stresses σ (x,t) is named linear elastic problem.
i
an an n
y ha

6.6.1 Governing Equations


le
liv or ec

The linear elastic problem is governed by the following equations:


M

.A

a) Cauchy’s equation (balance of linear momentum)


m

∂ 2 u (x,t)
uu
e

∇ · σ (x,t) + ρ0 b (x,t) = ρ0
X Th

∂t 2
er
tin

(3 equations) (6.42)
∂ σi j ∂ 2u j
+ ρ0 b j = ρ0 2 j ∈ {1, 2, 3}
on

.O

∂ xi ∂t
C

b) Constitutive equation (isotropic linear elastic)10


©

σ (x,t) = λ Tr (εε (x,t)) 1 + 2μεε (x,t)


(6 equations) (6.43)
σi j = λ δi j εll + 2μεi j i, j ∈ {1, 2, 3}

9 Here, linear elastic solid refers to a continuous medium constituted by a material that obeys
the linear elastic constitutive equation.
10 The symmetry of the stress and strain tensors entails that only six of the nine equations
are different from one another. In addition, when listing the unknowns, only the different
components of these tensors will be considered.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
278 C HAPTER 6. L INEAR E LASTICITY

c) Geometric equation (compatibility relation between infinitesimal strains and


displacements)

1
ε (x,t) = ∇S u (x,t) = (u ⊗ ∇ + ∇ ⊗ u)
2
(6 equations) (6.44)
1 ∂ ui ∂ u j
εi j = + i, j ∈ {1, 2, 3}
2 ∂ x j ∂ xi

These equations involve the following unknowns:

rs
• u (x,t) (3 unknowns)

ee
• ε (x,t) (6 unknowns) (6.45)
• σ (x,t) (6 unknowns)

s gin
and constitute a system of partial differential equations (PDEs). The system is

t d le En
composed of 15 differential equations with the 15 unknowns listed in (6.45).
These are of the type (•) (x, y, z,t), and, thus, must be solved in the R3 × R+

r
ba
ge ro or
eS m
space. The problem is well defined when the adequate boundary conditions are

ci
provided. f

ra
C d P cs
b
a
i
6.6.2 Boundary Conditions
an an n
y ha

6.6.2.1 Boundary Conditions in Space


le
liv or ec

Consider the boundary Γ ≡ ∂V of the solid is divided into three parts, Γu , Γσ


M

.A

and Γuσ , characterized by (see Figure 6.6)


 
Γu Γσ Γuσ = Γ ≡ ∂V ,
m

   (6.46)
uu
e

Γu Γσ = Γu Γuσ = Γuσ Γσ = {0}


/ .
X Th

er
tin

These allow defining the boundary conditions in space, that is, those conditions
on

.O

that affect the spatial arguments (x, y, z) of the unknowns (6.45) of the problem.
C

• Boundary Γu : prescribed displacements


©


u (x,t) = u∗ (x,t)
∀x ∈ Γu ∀t (6.47)
ui (x,t) = u∗i (x,t) i ∈ {1, 2, 3}

• Boundary Γσ : prescribed tractions



σ (x,t) · n = t∗ (x,t)
∀x ∈ Γσ ∀t (6.48)
σi j (x,t) · n j = ti∗ (x,t) i, j ∈ {1, 2, 3}

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
The Linear Elastic Problem 279

rs
ee
Figure 6.6: Boundary conditions in space.

s gin
t d le En
• Boundary Γuσ : prescribed displacements and tractions11

 

r
ui (x,t) = u∗i (x,t) 

ba


ge ro or
eS m
 i, j, k ∈ {1, 2, 3}, i =
j ∀x ∈ Γuσ ∀t (6.49)
σ jk (x,t) · nk = t ∗j (x,t) 

ci
f

ra
C d P cs
b
a
i
an an n

Example 6.2 – Exemplification of the boundary conditions in space.


y ha

le
liv or ec

Solution
M

.A

The different types of boundary conditions in space are illustrated in the fol-
lowing figure of a beam.
m

d
uu
e
X Th

er
tin
on

.O
C

11In Γuσ certain components (components i ) have prescribed displacements while the others
(components j ) have the traction vector prescribed.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
280 C HAPTER 6. L INEAR E LASTICITY

6.6.2.2 Boundary Conditions in Time: Initial Conditions


In general, at the initial or reference time t = 0 the displacements and velocities
are known.

u (x, 0) = 0 ⎪

 ∀x ∈ V

∂ u (x,t)  not . (6.50)

= u (x, 0) = v0 (x) ⎭
∂t 
t=0

rs
6.6.3 Quasi-Static Problem

ee
The system of equations (6.42) to (6.50) can be visualized, from a mechanical

s gin
point of view, as a system of actions or data (the body forces b (x,t), the traction
vector t∗ (x,t), the prescribed displacements u∗ (x,t) and the initial velocities

t d le En
v0 (x)) that, introduced into a mathematical model composed of the differen-

r
tial equations given in Section 6.6.1 and the boundary conditions described in

ba
ge ro or
eS m
Section 6.6.2, provides the response or solution in the form of the displacement

ci
f
field u (x,t), the deformation field ε (x,t) and the stress field σ (x,t).

ra
C d P cs
b
a
i

an an n

b (x,t) ⎪  ⎧
y ha


⎬ MAT HEMAT ICAL ⎨ u (x,t)
u∗ (x,t)
le
⇒ ⇒ ε (x,t)
liv or ec

∗ MODEL :
t (x,t) ⎪⎪ PDEs + BCs ⎩ (6.51)
⎭ σ (x,t)
M

.A

v0 (x)
  
  
m

not
Responses = R (x,t)
d

not
Actions = A (x,t)
uu
e
X Th

er
tin

In the most general case12 , both the actions and the responses depend on time
(see Figure 6.7) and the system of PDEs must be integrated over both the space
on

.O
C

Figure 6.7: Evolution of the response along time.

12 In this case (general problem), the problem is named dynamic problem.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
The Linear Elastic Problem 281

 
and the time variables R3 × R+ . However, in certain cases, the integration
space can be reduced in one dimension, the one corresponding to time. This is
the case for the so-called quasi-static problems.

Definition 6.3. A quasi-static linear elastic problem is a linear elas-


tic problem in which the acceleration is considered to be negligible,

∂ 2 u (x,t)
a= ≈0.
∂t 2

rs
This hypothesis is acceptable when the actions are applied slowly.

ee
In
 2such case, the variation of the actions A along time is slow

s gin
∂ A /∂t 2 ≈ 0 and, due to the continuous dependency of the results
on
 2the data, the variation of the response R along time is also small

t d le En
∂ R /∂t 2 ≈ 0 . Consequently, the second derivative of the response

r
is considered negligible and, in particular,

ba
ge ro or
eS m
ci
∂ 2 u (x,t)
f

ra
≈0.
C d P cs
∂t 2
b
a
i
an an n
y ha

le
liv or ec

The governing differential equations are reduced to the following in the case
of a quasi-static problem:
M

.A

a) Cauchy’s equation, also known as equilibrium equation.


m

d
uu
e

∂ 2 u (x,t)
X Th

∇ · σ (x,t) + ρ0 b (x,t) = ρ0 =0 (6.52)


er
tin

∂t 2
on

.O
C

b) Constitutive equation
©

σ (x,t) = λ Tr (εε (x,t)) 1 + 2μεε (x,t) (6.53)

c) Geometric equation, which no longer involves any time derivative.

1
ε (x,t) = ∇S u (x,t) = (u ⊗ ∇ + ∇ ⊗ u) (6.54)
2

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
282 C HAPTER 6. L INEAR E LASTICITY

The system of differential equations only needs to be integrated in space


(solved in R3 ) with the boundary conditions in space of Section 6.6.2.1. More-
over, time merely serves as a parameter describing the evolution of the actions,
which are usually described in terms of the load factor or pseudo-time λ (t).

⎫  ⎧
b (x, λ ) ⎬ MAT HEMAT ICAL ⎨ u (x, λ )
u∗ (x, λ ) ⇒ MODEL : ⇒ ε (x, λ )
⎭ PDEs + BCs ⎩ (6.55)
t∗ (x, λ ) σ (x, λ )
     

rs
not not
Actions = A (x, λ ) Responses = R (x, λ )

ee
s gin
In other words, for each value of the actions (characterized by a fixed value of
λ ∗ ), A (xλ ∗ ), a response R (x, λ ∗ ) is obtained. Varying the value of λ ∗ produces

t d le En
a family of actions and its corresponding family of responses.

r
ba
ge ro or
eS m
Example 6.3 – Application to a typical problem of strength of materials.

ci
f

ra
C d P cs
b
a
Solution
i
an an n

Consider a cantilever beam subjected to a force F (t) at its free end. Under
y ha

the quasi-static problem hypothesis, and considering a parametrized action


le
liv or ec

of the type λ F ∗ , the response (deflection at its free end) can be computed as
M

.A

F ∗l3
δ (λ ) = λ .
m

3EI
d
uu
e

This is the classical solution obtained in strength of materials.


X Th

er

Now, if the evolution along time of λ (t) can take any form, the value of
tin

δ (t) = δ (λ (t)) corresponding to each instant of time only depends on the


on

.O

corresponding value of λ .
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Solution to the Linear Elastic Problem 283

rs
ee
s gin
6.7 Solution to the Linear Elastic Problem

t d le En

r
The linear elastic problem can be typically solved following two different ap-

ba
ge ro or
eS m
proaches:

ci
f

ra
a) Displacement formulation
C d P cs
b
a
b) Stress formulation
i
an an n

Their names are directly related to which is the main unknown being considered
y ha

in each formulation (displacements or stresses, respectively).


le
liv or ec
M

.A

Remark 6.9. At present, the displacement formulation has greater


m

application because most numerical methods used to solve the linear


d
uu

elastic problem are based on this approach.


e
X Th

er
tin
on

.O

6.7.1 Displacement Formulation: Navier’s Equation


C

Consider the equations that constitute the linear elastic problem:

∂ 2u
∇ · σ + ρ0 b = ρ0 Cauchy’s equation
∂t 2
σ = λ Tr (εε ) 1 + 2μεε Constitutive equation (6.56)

1
ε = ∇S u = (u ⊗ ∇ + ∇ ⊗ u) Geometric equation
2

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
284 C HAPTER 6. L INEAR E LASTICITY


Γu : u = u∗
Boundary conditions in space (6.57)
Γσ : t∗ = σ · n


u (x, 0) = 0
. Initial conditions (6.58)
u (x, 0) = v0

The aim is to pose a reduced system in which only the displacement field
u (x,t) intervenes as an unknown. The first step consists in replacing the consti-

rs
tutive equation in the Cauchy’s equation, both given in (6.56).

ee
 

s gin
∂ 2u
∇ · σ + ρ0 b = ∇ · λ Tr (εε ) 1 + 2μεε + ρ0 b = ρ0 2
∂t

t d le En
(6.59)
  ∂ 2u

r
=⇒ λ ∇ · Tr (εε ) 1 + 2μ∇ · ε + ρ0 b = ρ0 2

ba
ge ro or
∂t

eS m
ci
f

ra
Consider the following identities13 .
C d P cs
b
a
i

an an n

  ∂ εi j ∂1 ∂ ui ∂ u j 1 ∂ 2 ui 1 ∂ ∂uj
y ha

∇·ε = = + = + =
i ∂xj ∂xj
2 ∂ x j ∂ xi 2 ∂ x j ∂ x j 2 ∂ xi ∂ x j
le
liv or ec

! "
1  1 ∂ 1 1
M

.A

= ∇2 u i + (∇ · u) = ∇2 u + ∇ (∇ · u) i ∈ {1, 2, 3}
2 2 ∂ xi 2 2 i
m

1 1
uu

∇ · ε = ∇ (∇ · u) + ∇2 u
e
X Th

2 2
er
tin

(6.60)
on

.O


   ∂ ∂ ∂ ul ∂ ∂ ul
C

∇ · Tr (εε ) 1 = (εll δi j ) = δi j = =
©

i ∂xj ∂ x j ∂ xl ∂ xi ∂ xl
∂  
= (∇ · u) = ∇ (∇ · u) i i ∈ {1, 2, 3}
∂ xi
 
∇ · Tr (εε ) 1 = ∇ (∇ · u)
(6.61)

  de f
13 The Laplace operator of a vector v is defined as ∇2 v i = ∂ 2 vi /(∂ x j ∂ x j ).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Solution to the Linear Elastic Problem 285

Equation (6.59) can be rewritten by replacing the expressions in the identities


(6.60) and (6.61), resulting in

⎨ ∂ 2u
Navier’s (λ + μ) ∇ (∇ · u) + μ∇2 u + ρ0 b = ρ0 2
equation ⎩ ∂t (6.62)
(λ + μ) u j, ji + μ ui, j j + ρ0 bi = ρ0 üi i ∈ {1, 2, 3}

which constitutes a system of second-order PDEs in displacements u (x,t) (that


must be, thus, integrated in R3 × R+ ), and receives the name of Navier’s equa-
tion.

rs
The boundary conditions can also be written in terms of the displacements as

ee
follows. Replacing the constitutive equation of (6.56) in the boundary conditions
in Γσ of (6.57) results in

s gin
 
t∗ = σ · n = λ Tr (εε ) 1 + 2μεε · n = λ (Tr (εε )) n + 2μεε · n =

t d le En
  (6.63)

r
= λ (∇ · u) n + 2μ ∇S · u · n = λ (∇ · u) n + μ (u ⊗ ∇ + ∇ ⊗ u) · n

ba
ge ro or
eS m
ci
f

ra
and the boundary conditions in space (6.57) expressed in terms of the displace-
C d P cs
b
a
ments are obtained.
i
an an n


u = u∗
y ha

in Γu
ui = u∗i
le
i ∈ {1, 2, 3}
liv or ec

 (6.64)
λ (∇ · u) n + μ (u ⊗ ∇ + ∇ ⊗ u) · n = t∗
M

.A

in Γ σ
λ ul,l ni + μ (ui, j n j + u j,i n j ) = ti∗ i, j ∈ {1, 2, 3}
m

d
uu

The initial conditions (6.58) remain unchanged. Integrating the system (6.62)
e
X Th

yields the displacement field u (x,t). Differentiation of this field and substitution
er
tin

in the geometric equation of (6.56) produces the strain field ε (x,t), and, finally,
replacing the strain in the constitutive equation results in the stress field σ (x,t).
on

.O
C

6.7.1.1 Navier’s Equation in Cylindrical and Spherical Coordinates


Navier’s equation (6.62) is expressed in compact or index notation and is inde-
pendent of the coordinate system considered. The components of this equation
are expressed as follows in the cylindrical and spherical coordinate systems (see
section 2.15).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
286 C HAPTER 6. L INEAR E LASTICITY

Cylindrical coordinates

∂ e 2μ ∂ ωz ∂ ωθ ∂ 2 ur
(λ + 2μ) − + 2μ + ρbr = ρ 2
∂r r ∂θ ∂z ∂t
1 ∂e ∂ ωr ∂ ωz ∂ 2 uθ (6.65)
(λ + 2μ) − 2μ + 2μ + ρbθ = ρ 2
r ∂θ ∂z ∂r ∂t
∂ e 2μ ∂ (rωθ ) 2μ ∂ ωr ∂ 2 uz
(λ + 2μ) − + + ρbz = ρ 2
∂z r ∂r r ∂θ ∂t

rs
where

ee

1 1 ∂ uz ∂ uθ

s gin
ωr = −Ωθ z = −
2 r ∂θ ∂z

t d le En
1 ∂ ur ∂ uz
ωθ = −Ωzr = −

r
2 ∂z ∂r

ba
ge ro or
eS m

ci
1 1 ∂ (ruθ ) 1 ∂ ur
ωz = −Ωrθ = − f

ra
C d P cs
2 r ∂r r ∂θ
b
a
i
an an n

1 ∂ (rur ) 1 ∂ uθ ∂ uz
e= + +
y ha

r ∂r r ∂θ ∂z
le
liv or ec
M

.A

⎡ ⎤
x = r cos θ
x (r, θ , z) ≡ ⎣ y = r sin θ ⎦
m

not
d

z=z
uu
e
X Th

er
tin
on

.O
C

Figure 6.8: Cylindrical coordinates.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Solution to the Linear Elastic Problem 287

Spherical coordinates

 
∂e 2μ ∂ ωφ sin θ 2μ ∂ ωθ ∂ 2 ur
(λ + 2μ) − + + ρbr = ρ 2
∂ r r sin θ ∂θ r sin θ ∂ φ ∂t
 
1 ∂e 2μ ∂ ωr 2μ ∂ rωφ sin θ ∂ 2 uθ (6.66)
(λ + 2μ) − + + ρbθ = ρ 2
r ∂ θ r sin θ ∂ φ r sin θ ∂r ∂t
1 ∂ e 2μ ∂ (rωθ ) 2μ ∂ ωr ∂ 2 uφ
(λ + 2μ) − + + ρbφ = ρ 2
r sin θ ∂ φ ∂r r ∂θ ∂t

rs
r

ee
where
'   (

s gin
1 1 ∂ uφ sin θ 1 ∂ uθ
ωr = −Ωθ φ = −
r sin θ ∂θ r sin θ ∂ φ

t d le En
2
'  (

r
1 ∂ ur 1 ∂ ruφ

ba
1

ge ro or
ωθ = −Ωφ r = −

eS m
2 r sin θ ∂ φ r ∂r

ci
f

ra

C d P cs
b
a
1 1 ∂ (ruθ ) 1 ∂ ur
ωz = −Ωrθ = −
i
an an n

2 r ∂r r ∂θ
'   (
y ha


1 ∂ r2 ur sin θ ∂ (ruθ sin θ ) ∂ ruφ
le
e= 2 + +
liv or ec

r sin θ ∂r ∂θ ∂φ
M

.A
m

⎡ ⎤
d

x = r sin θ cos φ
uu
e

x (r, θ , φ ) ≡ ⎣ y = r sin θ sin φ ⎦


not
X Th

er
tin

z = z cos θ
on

.O
C

Figure 6.9: Spherical coordinates.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
288 C HAPTER 6. L INEAR E LASTICITY

6.7.2 Stress Formulation: Beltrami-Michell Equation


This formulation is solely valid for the quasi-static case discussed in Sec-
tion 6.6.3. Consider, thus, the equations that constitute the quasi-static linear
elastic problem:

∇ · σ + ρ0 b = 0 Equilibrium equation
ν 1+ν
ε = − Tr (σ σ)1+ σ Inverse constitutive equation (6.67)
E E
1
ε = ∇S u = (u ⊗ ∇ + ∇ ⊗ u) Geometric equation

rs
2

ee


s gin
Γu : u = u∗
Boundary conditions in space (6.68)
Γσ : t∗ = σ · n

t d le En

r
ba
ge ro or
where the inverse constitutive (6.24) (strains in terms of stresses) has been con-

eS m
ci
sidered in (6.67).
f

ra
The starting point of the stress formulation is the geometric equation of (6.67)
C d P cs
b
a
from which, by means of successive differentiation, the displacements are elim-
i
an an n

inated and the compatibility equations14 are obtained,


y ha

εi j, kl + εkl, i j − εik, jl − ε jl, ik = 0 i, j, k, l ∈ {1, 2, 3} .


le
(6.69)
liv or ec
M

.A

Then, the equations of the problem are deduced in the following manner:
a) The constitutive equation of (6.67) is replaced in the compatibility equa-
m

tions (6.69).
uu
e

b) The resulting expression is introduced in the equilibrium equation of (6.67).


X Th

er
tin

This results in the equation


on

.O

Beltrami-Michell equation
C

ν
©

1
∇2 σi j + σll,i j = − δi j (ρ0 bl ), l − (ρ0 bi ), j − (ρ0 b j ), i (6.70)
1+ν 1−ν
i, j ∈ {1, 2, 3}

which receives the name of Beltrami-Michell equation and constitutes a system


of second-order PDEs in stresses σ (x) that must be solved in R3 .
The boundary conditions of this system are the equilibrium equation of (6.67),
which, being a system of first-order PDEs, acts as the boundary conditions of the
second-order system in (6.70), and the boundary conditions in Γσ .
14 The deduction of the compatibility equations has been studied in Chapter 3, Section 3.3.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Unicity of the Solution to the Linear Elastic Problem 289

∇ · σ + ρ0 b = 0 Equilibrium equation (6.71)

σ · n = t∗ in Γσ Boundary conditions in Γσ (6.72)

The integration of the system in (6.70) yields the stress field σ (x). Substi-
tution of the stresses in the inverse constitutive equation of (6.67) results in
the strains ε (x). However, to obtain the displacement field u (x), the geomet-
ric equations must be integrated, taking into account the boundary conditions

rs
in Γu 15 . ⎧

ee
⎪  
⎨ ε (x) = 1 u (x) ⊗ ∇ + ∇ ⊗ u (x) x∈V

s gin
2 (6.73)

⎩ u (x) = u∗ (x) ∀x ∈ Γu

t d le En

r
Thus, the system of second-order PDEs must be integrated in R3 .

ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
Remark 6.10. The need to integrate the second system (6.73) (when
i
an an n

the stress formulation is followed) is a disadvantage (with respect


y ha

to the displacement formulation described in Section 6.7.1) when


le
numerical methods are used to solve the linear elastic problem.
liv or ec
M

.A
m

6.8 Unicity of the Solution to the Linear Elastic Problem


uu
e
X Th

er
tin
on

Theorem 6.1. The solution


.O

⎡ ⎤
u (x,t)
C

R (x,t) ≡ ⎣ ε (x,t) ⎦
not
©

σ (x,t)
to the linear elastic problem posed in (6.42) to (6.44) is unique.

Proof
Consider the actions defined by A (x,t) ≡ [b (x,t) , u∗ (x,t) , t∗(x,t) , v0 (x)]T ,
not

in the domains V , Γu , Γσ and V , respectively, (satisfying Γσ Γu = ∂V and


15An analytical procedure to integrate these geometric equations was provided in Chapter 3,
Section 3.4.2

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
290 C HAPTER 6. L INEAR E LASTICITY

Figure 6.10: Linear elastic problem.

rs
ee

Γσ Γu = 0) / act on the linear elastic problem schematically represented in Fig-

s gin
ure 6.10.
The possible solutions R (x,t) ≡ [u (x,t) , ε (x,t) , σ (x,t)]T to the linear elas-
not

t d le En
tic problem must satisfy the equations:

r
ba
ge ro or
∂ 2u

eS m
∇ · σ + ρ0 b = ρ0 Cauchy’s equation

ci
∂t 2
f

ra
C d P cs
σ = λ Tr (εε ) 1 + 2μεε (6.74)
b
a
Constitutive equation
i
an an n

1
ε = ∇S u = (u ⊗ ∇ + ∇ ⊗ u) Geometric equation
y ha

2
le
liv or ec


M

Γu : u = u∗
.A

Boundary conditions in space (6.75)


Γσ : t∗ = σ · n
m

d
uu
e


X Th

er

u (x, 0) = 0
tin

. Initial conditions (6.76)


u (x, 0) = v0
on

.O
C

The unicity of the solution is proven as follows. Suppose the solution is not
©

unique, that is, there exist two different solutions to the problem,
⎡ (1) ⎤ ⎡ (2) ⎤
u (x,t) u (x,t)
not ⎢ ⎥ not ⎢ ⎥
R (1) (x,t) ≡ ⎣ ε (1) (x,t) ⎦ and R (2) (x,t) ≡ ⎣ ε (2) (x,t) ⎦
σ (1) (x,t) σ (2) (x,t) (6.77)

such that R (1) = R (2) ,

which, therefore, must satisfy equations (6.74) to (6.76) and are the elastic re-
sponses to the same action A (x,t) ≡ [b (x,t) , u∗ (x,t) , t∗ (x,t) , v0 (x)]T . Con-
not

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Unicity of the Solution to the Linear Elastic Problem 291

sider now a possible response constituted by the difference R (2) −R R(1) ,


⎡ (2) ⎤ ⎡ ⎤
u (x,t) − u(1) (x,t) + (x,t)
u
+ (x,t) de f not ⎢ ⎥ de f ⎢ ⎥
R = R(2) −RR(1) ≡ ⎣ ε (2) (x,t) − ε (1) (x,t) ⎦ = ⎣ ε+ (x,t) ⎦ . (6.78)
σ (2) (x,t) − σ (1) (x,t) + (x,t)
σ

+ satisfies the following equations:


Note how the answer R
• Cauchy’s equation with b = 0 16

rs
 
+ (x,t) = ∇ · σ (2) (x,t) − σ (1) (x,t) = ∇ · σ (2) − ∇ · σ (1) =
∇·σ

ee
s gin
' ( ' (
∂ 2 u(2) ∂ 2 u(1) (6.79)
= −ρ0 b + ρ0 − −ρ0 b + ρ0 =

t d le En
∂t 2 ∂t 2

r
ba
ge ro or
∂ 2 u(2) ∂ 2 u(1) +
∂ 2u

eS m
= ρ0 − ρ = ρ

ci
0 0
∂t 2 ∂t 2
f ∂t 2

ra
C d P cs
b
a
• Constitutive equation17
i
an an n
y ha

+ (x,t) = σ (2) (x,t) − σ (1) (x,t) = C : ε (2) −C


σ C : ε (1) =
le
 
liv or ec

(6.80)
= C : ε (2) − ε (1) = C : ε+
M

.A
m

• Geometric equation
d
uu
e
X Th

ε+ (x,t) = ε (2) (x,t) − ε (1) (x,t) = ∇S u(2) − ∇S u(1) =


er
tin

  (6.81)
= ∇S u(2) − u(1) = ∇S u +
on

.O
C

• Boundary conditions in Γu with u +∗ = 0





⎨u+ (x,t) = u(2) (x,t) − u(1) (x,t) = u∗ − u∗ = 0 ∀t =⇒
Γu → (6.82)
⎪ + (x,t) .
⎩ =⇒ ∂ u

=u + (x,t) = 0
∂t
16 The fact that the Nabla operator (∇ ∗ (•)) is a linear operator is used advantageously here,
that is, ∇ ∗ (a + b) = ∇ ∗ a + ∇ ∗ b, where ∗ symbolizes any type of differential operation.
Likewise, the operator ∂ 2 (•,t)/∂t 2 is also a linear operator.
17 The property that the operator C : is a linear operator is applied here, that is,
C : (a + b) = C : a +CC : b.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
292 C HAPTER 6. L INEAR E LASTICITY

• Boundary conditions in Γσ with +t∗ = 0


 
+ (x,t) · n = σ (2) (x,t) − σ (1) (x,t) · n = σ (2) · n − σ (1) · n =
Γσ → σ
(6.83)
= t∗ − t∗ =0

• Initial conditions with v0 = 0



rs

⎨u + (x, 0) = u(2) (x, 0) − u(1) (x, 0) = 0 − 0 = 0

ee
(6.84)
⎪ + (x, 0) .
⎩ ∂u

s gin
. .
+ (x, 0) = u(2) (x, 0) − u(1) (x, 0) = v0 − v0 = 0
=u
∂t

t d le En
Consider now the calculation of the integral

r
ba
ge ro or
eS m
= 0 in Γσ Divergence

ci
  .     Theorem  .
. f

ra
+ ·u + · u + ·u
C d P cs
n· σ + dS = n·σ + dS = ∇· σ + dV = 0 , (6.85)
b
a
 
i
∂V Γu Γσ
an an n

= 0 in Γu V
y ha

le
where the conditions (6.82) and (6.83) have been applied. Operating on (6.85)
liv or ec

results in

M

.A


⎪  .   .  . + .  . T

⎪ ∂ 2u
⎨∇· σ + ·u + = ∇·σ + ·u ++σ + : ∇u + = ρ0 2 · u ++σ + : ∇u +
m

∂t
d
uu

⎪  .  ∂ σ+ . . .
e


⎪ ∂ ∂ +
u ∂ +
2u . ∂ +
u

X Th

i j j j j
⎩ σ+i j u+j = u+j + σ+i j = ρ0 2 u+j + σ+ ji i, j ∈ {1, 2, 3}
er
tin

∂ xi ∂ xi ∂ xi ∂t ∂ xi
(6.86)
on

.O

where the condition (6.79) has been considered. On the other hand18 ,
C

 . T . 1 . . 1 . . . .
©

∇u+ =u +⊗∇ = +⊗∇+∇⊗u


u + + u +⊗∇−∇⊗u + =⇒
+ = ε+ + Ω
2   2   
. . . .
ε+ = ∇ u
S
+ + =∇ u
Ω a
+
 . T . .  . T .
+ : ∇u
σ + =σ + : ε+ + σ
+ :Ω + =⇒ + : ∇u
σ + =σ + : ε+ .
  
=0
(6.87)
.
18The fact that σ + is an antisymmetric one is considered here,
+ is a symmetric tensor and Ω
. .
which leads to σ + =σ
+ :Ω + i j = 0.
+i j Ω

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Unicity of the Solution to the Linear Elastic Problem 293

In addition19 , +
v2
. .
.   
∂2u + . ∂2u+ . 1 1 ∂ + +·u
∂ u
v·++
v
ρ0 2 · u+ = ρ0 2 · u+ = ρ0 = ρ0 =
∂t ∂t 2 ∂t 2 ∂t (6.88)

d 1 2 ∂2u+ . d 1 2
= ρ0 +
v =⇒ + = ρ0
ρ0 2 · u +
v .
dt 2 ∂t dt 2

Replacing (6.88) and (6.87) in (6.86), and the resulting expression in (6.85), and
taking into account the definition of internal energy U given in (6.10) produces

rs
  .   .
d 1 2

ee
+
∇· σ ·u + dV = ρ0 +
v dV + σ + : ε+ dV = 0 =⇒
dt 2

s gin
V  V
.  V
d 1
ρ0 +
v2 dV + σ+ : ε+ dV = 0 =⇒ (6.89)

t d le En
dt 2
V V
     

r
ba
ge ro or
+ /dt d U+/dt

eS m
dK

ci
f

ra
C d P cs
d  + +
+ d U+
b
a
dK
i
+ = K + U = 0 ∀t ≥ 0 . (6.90)
an an n

dt dt dt
y ha

le
liv or ec

Note, though, that at the initial time t = 0 the following is satisfied (see (6.10),
M

.A

(6.13) and (6.84))


    ⎫
m

 1  1 ⎪
+ = v0 dV = 0 ⎪
d

K ρ0+ v  dV =
2
ρ0 + v0 · + ⎪

uu

2 2  ⎪

e

t=0 t=0
. ⎪
⎬  
X Th

V V
er

+0 = 0 
tin

    
u
 ⇒ +
K + +
U  =0
+   1   ⎪
⎪ t=0
U  = û (x,t)  dV = ε+ : C : ε+ dV = 0 ⎪
on


.O

t=0 t=0 2  
t=0 t=0 ⎪

V V  ⎪

C

=0
©

(6.91)
and the integration of (6.90) with the initial condition (6.91) leads to

+ + U+ = 0
K ∀t ≥ 0 , (6.92)

where 
+= 1
K ρ0 +
v2 dV ≥ 0 ∀t ≥ 0 . (6.93)
2 
V
≥0

de f
19 v| = +
Here, the definition |+ v is used.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
294 C HAPTER 6. L INEAR E LASTICITY

Comparing (6.92) and (6.93) necessarily leads to the conclusion


 
+ + U+ = 0
K 1
∀t ≥ 0 =⇒ U+ = ε+ : C : ε+ dV ≤ 0 ∀t ≥ 0 . (6.94)
+
K≥0 2
V

On the other hand, since the tensor of elastic constants C is positive-definite


(see (6.32)) ,

ε+ (x,t) : C : ε+ (x,t) ≥ 0 ∀x ∈ V ∀t ≥ 0 =⇒


rs
1 (6.95)
U+ = ε+ : C : ε+ dV ≥ 0 ∀t ≥ 0 .

ee
2
V

s gin
Then, comparing (6.94) and (6.95) necessarily leads to

t d le En
 
U+ (t) ≤ 0 1

r
+
∀t ≥ 0 =⇒ U (t) = +
ε :C:+ ε dV = 0 ∀t ≥ 0 . (6.96)

ba
ge ro or
+

eS m
U (t) ≥ 0 2

ci
fV

ra
C d P cs
b
a
Considering once more the positive-definite condition of tensor C 20 ,
i
an an n


y ha

1
U+ = ε+ : C : ε+ dV = 0 ∀t ≥ 0 =⇒ ε+ : C : ε+ = 0 ∀x , ∀t ≥ 0 (6.97)
le
2   
liv or ec

V ≥0
M

.A

and, necessarily, from the positive-definite condition of C it is deduced that


m

ε+ : C : ε+ = 0 ⇐⇒ ε+ (x,t) = 0 ∀x , ∀t ≥ 0
uu

(6.98)
e
X Th

er
tin

ε+ (x,t) = ε (2) − ε (1) = 0 =⇒ ε (2) = ε (1) . (6.99)


on

.O

In addition, replacing (6.99) in (6.81) results in


C


1 ∂ u+i ∂ u+j
©

ε+ (x,t) = ∇ · u
S
+ = 0 =⇒ + =0 i, j ∈ {1, 2, 3} , (6.100)
2 ∂ x j ∂ xi

20 The following theorem of integral calculus is applied here:



If φ (x) ≥ 0 and φ (x) dΩ = 0 =⇒ φ (x) = 0 ∀x ∈ Ω .
Ω

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Saint-Venant’s Principle 295

which is a system of six homogeneous and first-order PDEs. Its integration leads
to the solution21

+ (x,t) = Ω
u + · x + +c with
 
rotation translation

⎡ ⎤ ⎡ ⎤ (6.101)
0 −θ+3 θ+2 c+1
+ not ⎢ ⎥ not ⎢ ⎥
Ω ≡ ⎣ θ+3 0 −θ+1 ⎦ and +c ≡ ⎣ c+2 ⎦ ,
−θ+2 θ+1 c+3

rs
0

ee
where Ω + is an antisymmetric tensor (rotation tensor dependent on three con-

s gin
stants θ+1 , θ+2 and θ+3 ) and +c is a constant vector equivalent to a translation. Ulti-
mately, the solution (6.100) to the system (6.101) are the displacements u + (x,t)

t d le En
compatible with a null strain ε+ (x,t) = 0, which correspond to a rigid body

r
+ and +c are determined by imposing the

ba
motion. The integration constants in Ω

ge ro or
eS m
boundary conditions (6.82) (+ u (x,t) = 0 ∀x ∈ Γu ), therefore, if the rigid body

ci
f

ra
motion is impeded through the restrictions in Γu , one obtains Ω + = 0 and +c = 0.
C d P cs
b
a
In conclusion,
i
an an n


y ha

+ (x,t) = Ω
u + · x ++c
=⇒ u + (x,t) = u(2) − u(1) = 0 =⇒ u(2) = u(1) .
le
liv or ec

+
Ω ≡ 0 ; +c ≡ 0
M

.A

(6.102)
Finally, replacing (6.99) in (6.80) yields
m

d
uu

+ (x,t) = C : ε+ = 0 = σ (2) − σ (1) σ (2) = σ (1) .


e

σ =⇒ (6.103)
X Th

er
tin

Then, observing (6.99) , (6.102) and (6.103) leads to the conclusion


on

.O


u(2) = u(1) ⎬
C

ε (2) = ε (1) =⇒ R (2) = R (1) . (6.104)


©


σ (2) = σ (1)

Therefore, the solution is unique (QED).

6.9 Saint-Venant’s Principle


Saint-Venant’s principle is an empirical principle that does not have a rigor-
ous proof. Consider a solid Ω that is subjected to a system of forces on its

21 This solution can be obtained applying the methodology used in Chapter 3, Section 3.4.2
to integrate the strain field.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
296 C HAPTER 6. L INEAR E LASTICITY

rs
ee
s gin
Figure 6.11: Saint-Venant’s principle.

t d le En

r
ba
ge ro or
eS m
boundary characterized by the traction vector t∗ (see Figure 6.11). These ac-

ci
tions will lead f
to a solution or response in displacements, strains and stresses,

ra
not  T
C d P cs
R (I) (x,t) ≡ u(I) (x,t), ε (I) (x,t) , σ (I) (x,t) . Consider now a part Γ̂ of the
b
a
i
boundary Γσ Γ̂ ⊂ Γσ of said medium, whose typical dimension is , and re-
an an n
y ha

place the system of actions applied on the boundary, t(I) , by another system,
le
t(II) , that is statically equivalent to t(I) 22 , without modifying the actions on the
liv or ec

rest of Γσ . Modifying the actions in this way will presumably result in the new
M

not  T
.A

responses R (II) (x,t) ≡ u(II) (x,t) , ε (II) (x,t) , σ (II) (x,t) .


m

Saint-Venant’s principle states that, for the points belonging to the domain Ω
d

that are sufficiently far from the boundary Γ̂ , the solution in both cases is prac-
uu
e

tically the same, that is, for a point P of the interior of Ω ,


X Th

er
tin


u(I) (x p ,t) ≈ u(II) (x p ,t) ⎪
on

.O

⎬ 
ε (I) (x p ,t) ≈ ε (II) (x p ,t) ∀P  δ   . (6.105)
C



©

σ (I) (x p ,t) ≈ σ (II) (x p ,t)

In other words, if the distance δ between the point being considered and the
part of the boundary in which the actions have been modified is large in com-
parison with the dimension  of the modified zone, the response in said point is
equivalent in both cases.

22 Two systems of forces t(I) and t(II) are said to be statically equivalent if the resultant
(forces and moments) of both systems is the same.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Linear Thermoelasticity. Thermal Stresses and Strains 297

Example 6.4 – Description of Saint-Venant’s principle in strength of materi-


als and how it relates to the concept of stress.

Solution
Consider a beam (or prismatic piece) with a cross-section A subjected to
a tensile point force F in its ends, as shown in the figure below. The exact
solution to the original elastic problem (system (I)) is extremely complicated,
especially in the vicinity of the points of application of the point forces. If
the forces F are now replaced by a statically equivalent system of uniformly
distributed tensile loads in the end sections σ = F/A (system (II)), the elastic

rs
solution to the corresponding problem is extremely simple and coincides (for

ee
a Poisson’s ratio of ν = 0) with the axial stress solution provided by strength

s gin
of materials (uniformly distributed stresses in all the piece, σx = F/A). At
a far enough distance from the beam’s ends (once or twice the edge), Saint-

t d le En
Venant’s principle allows approximating solution (I) with solution (II), and
also allows dimensioning the strength characteristics of the piece for practical

r
ba
ge ro or
eS m
purposes.

ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

6.10 Linear Thermoelasticity. Thermal Stresses and


Strains
The main difference of linear thermoelasticity with respect to the linear elasticity
studied up to this point is that the deformation process is no longer assumed to
be isothermal (see Section 6.1). Now, the thermal effects are included and the

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
298 C HAPTER 6. L INEAR E LASTICITY

temperature θ (x,t) is considered to evolve along time, that is,

not
θ (x,t) = θ (x, 0) = θ0 ,
. ∂ θ (x,t) (6.106)
θ (x,t) = = 0 .
∂t
Nevertheless, the hypothesis that the processes are adiabatic (slow) is maintained
and, thus,
ρ0 r − ∇ · q ≈ 0 . (6.107)

rs
6.10.1 Linear Thermoelastic Constitutive Equation

ee
s gin
Hooke’s law (6.6) in this case is generalized to

t d le En
σ = C : ε − β (θ − θ0 )
, (6.108)

r
ba
σi j = Ci jkl εkl − βi j (θ − θ0 ) i, j ∈ {1, 2, 3}

ge ro or
eS m
ci
f

ra
Here, C is the tensor of elastic constants defined in (6.7), θ (x,t) is the tempera-
C d P cs
b
a
ture field, θ0 (x) = θ (x, 0) is the distribution of temperatures in the neutral state
i
an an n

(reference configuration) and β is the (symmetric) tensor of thermal properties.


y ha


le
liv or ec

Tensor of thermal β = βT
(6.109)
M

.A

properties βi j = β ji i, j ∈ {1, 2, 3}
m

In the case of an isotropic material, tensor C must be a fourth-order isotropic


uu
e

tensor and β , a second-order isotropic one23 , that is,


X Th

er
tin


C = λ 1 ⊗ 1 + 2μI
on

.O

 
Ci jkl = λ δi j δkl + μ δik δ jl + δil δ jk i, j, k, l ∈ {1, 2, 3}
C

 (6.110)
©

β = β1
β i j = β δi j i, j ∈ {1, 2, 3}

where now a single thermal property β appears in addition to the elastic con-
stants λ and μ. Replacing (6.110) in the constitutive equation (6.108) and defin-
not
ing (θ − θ0 ) = Δ θ , yields

23 The most general expression of a second-order isotropic tensor is β = β 1 ∀β .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Linear Thermoelasticity. Thermal Stresses and Strains 299

Constitutive equation of an
isotropic linear thermoelastic material
(6.111)
σ = λ Tr (εε ) 1 + 2μεε − β Δ θ 1
σi j = λ εll δi j + 2μεi j − β Δ θ δi j i, j ∈ {1, 2, 3}

6.10.2 Inverse Constitutive Equation

rs
Equation (6.111) can be inverted as follows.

ee

s gin
−1 −1

⎨σ = C : ε −Δθ β ⇒ ε = C : σ +Δθ C : β = C −1 : σ + Δ θ α
  
α

t d le En

⎩ de f −1
α = C : β → Tensor of thermal expansion coefficients

r
ba
ge ro or
(6.112)

eS m
ci
where α is a second-order (symmetric) tensor involving six thermal properties
f

ra
C d P cs
named coefficients of thermal expansion. For an isotropic case, in agreement
b
a
with (6.111) and (6.24), and after certain algebraic manipulation, one obtains
i
an an n
y ha

le
Inverse constitutive equation of an
liv or ec

isotropic linear thermoelastic material


M

.A

ν 1+ν
ε =− σ)1+
Tr (σ σ + αΔ θ 1 (6.113)
m

E E
d
uu

ν 1+ν
e

εi j = − σll δi j + σi j + αΔ θ δi j i, j ∈ {1, 2, 3}
X Th

er
tin

E E
on

.O

Here, α is a scalar denoted as coefficient of thermal expansion, related to the


C

thermal property β in (6.111) by means of


©

Thermal expansion → α = 1 − 2ν β . (6.114)


coefficient E

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
300 C HAPTER 6. L INEAR E LASTICITY

6.10.3 Thermal Stresses and Strains


Comparing the linear elastic constitutive equation (6.20) and the linear thermoe-
lastic one (6.111) suggests the following decomposition.

σ = λ Tr (εε ) 1 + 2μεε − β Δ θ 1 = σ nt − σ t
     
σ nt σt
⎧ (6.115)

⎨ Non-thermal stress → σ nt de f
= λ Tr (εε ) 1 + 2μεε

rs

⎩ Thermal stress de f
→ σt = β Δθ 1

ee
s gin
Here, σ nt represents the stress produced if there do not exist any thermal phe-
nomena and σ t is named thermal stress and acts as the “correcting” stress due

t d le En
to the thermal increment.

r
A similar operation can be performed on the inverse constitutive equations

ba
ge ro or
eS m
for the linear elastic and linear thermoelastic cases of (6.24) and (6.113), re-

ci
spectively, resulting in
f

ra
C d P cs
b
a
ν 1+ν
i
σ)1+
ε = − Tr (σ σ + α Δ θ 1 = ε nt + ε t
an an n

 E  E   t 
y ha

ε nt ε
le
liv or ec

⎧ (6.116)
⎪ ν 1+ν
M

.A

⎨ Non-thermal strain → ε nt de f
σ)1+
= − Tr (σ σ
E E

m

⎩ Thermal strain de f
→ εt = α Δθ 1
d
uu
e
X Th

er
tin

In conclusion, the stress and strain tensors in linear thermoelasticity can be de-
composed into
on

.O

Non-thermal Thermal
C

Total component component


©

σ nt = C : ε σt = Δθ β
σ = σ nt − σ t Isotropic material: Isotropic material: (6.117)
σ nt = λ Tr (εε ) 1 + 2μεε σt = β Δθ 1

ε nt = C −1 : σ εt = Δθ α
ε = ε nt + ε t Isotropic material: Isotropic material: (6.118)
ν 1+ν εt = α Δθ 1
ε nt = − Tr (σ σ)1+ σ
E E

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Linear Thermoelasticity. Thermal Stresses and Strains 301

where the thermal components appear due to the thermal processes being taken
into account. The following expressions result from (6.117) and (6.118).
 
ε nt = C −1 : σ =⇒ σ = C : ε nt = C : ε − ε t (6.119)

 
σ nt = C : ε =⇒ ε = C −1 : σ nt = C −1 : σ + σ t (6.120)

rs
Remark 6.11. Unlike what occurs in elasticity, in the thermoelastic

ee
case a state of null strain in a point of a medium does not imply a
state of null stress in said point. In effect, for ε = 0 in (6.117),

s gin
σ t = −β Δ θ 1 = 0 .
ε = 0 =⇒ σ nt = 0 =⇒ σ = −σ

t d le En

r
ba
ge ro or
eS m
ci
f

ra
Δ θ = 0
C d P cs
σ t = −β Δ θ 1
σ = −σ
b
a
ε =0
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e

Remark 6.12. Analogously, in thermoelasticity a state of null stress


X Th

er
tin

in a point of a medium does not imply a state of null strain in said


point since (6.118) with σ = 0 yields
on

.O

σ = 0 =⇒ ε nt = 0 =⇒ ε = ε t = α Δ θ 1 = 0 .
C

Δ θ = 0
ε = εt = α Δθ 1
σ =0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
302 C HAPTER 6. L INEAR E LASTICITY

rs
ee
s gin
Figure 6.12: Actions on a continuous medium.

t d le En

r
ba
6.11 Thermal Analogies

ge ro or
eS m
ci
f

ra
The thermal analogies arise from the search of procedures to solve the linear
C d P cs
b
a
thermoelastic problem using the strategies and methodologies developed in Sec-
i
tion 6.7 for the linear elastic problem (without considering thermal effects).
an an n
y ha

Two analogies are presented in this section which, for the sake of simplicity,
are restricted to the isotropic quasi-static problem, although they can be directly
le
liv or ec

extrapolated to the general anisotropic dynamic problem.


M

.A

6.11.1 First Thermal Analogy (Duhamel-Newman Analogy)


m

Consider the continuous medium in Figure 6.12 on which the body forces b (x,t)
uu
e

and an increment of temperature Δ θ (x,t) are acting, and on whose boundaries


X Th

er

Γu and Γσ act the prescribed displacements u∗ (x,t) and a traction vector t∗ (x,t),
tin

respectively.
on

.O

The equations of the (isotropic quasi-static) linear thermoelastic problem are


C


©




⎪ ∇ · σ + ρ0 b = 0 Equilibrium equation

Governing
σ = C : ε −β Δθ 1 Constitutive equation
equations ⎪



⎩ ε = ∇S u Geometric equation (6.121)

Boundary Γu : u = u∗
conditions Γσ : σ · n = t∗

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Thermal Analogies 303

which compose the actions (data) A (x,t) and responses (unknowns) R (x,t) of
the problem24 .


b (x,t) ⎪  ⎧


⎬ MAT HEMAT ICAL ⎨ u (x,t)
u (x,t)
⇒ MODEL : ⇒ ε (x,t)
t∗ (x,t) ⎪
⎪ + ⎩ (6.122)
⎭ PDEs BCs σ (x,t)
Δ θ (x,t)
  
  
Responses = R (I) (x,t)
not
Actions = A (I) (x,t)
not

rs
ee
To be able to apply the resolution methods typical of the liner elastic problem

s gin
developed in Section 6.7, the thermal term in the equations of the thermoelastic
problem (6.121) must be eliminated (at least, in appearance). To this aim, the

t d le En
σ t is replaced in (6.121) as follows.
decomposition of the stress tensor σ = σ nt −σ

r
ba
a) Equilibrium equation

ge ro or
eS m
ci
f

ra
σ = σ nt − σ t =⇒
C d P cs
b
a
∇·σ = ∇ · σ nt − ∇ · σ = ∇ · σ nt − ∇ (β Δ θ )
t
i
(6.123)
an an n


y ha

β Δθ1
le
 
liv or ec

1
∇ · σ + ρ0 b = 0 =⇒ ∇ · σ nt + ρ0 b − ∇ (β Δ θ ) = 0
M

.A

ρ0
  
(6.124)
m

not
= b̂
d
uu
e

=⇒ ∇ · σ + b̂ = 0
nt
X Th

er
tin

which constitutes the equilibrium equation of the medium subjected to the


on

.O

pseudo-body forces b̂ (x,t) defined by



C



©

⎪ 1
⎨ b̂ (x,t) = b (x,t) − ∇ (β Δ θ )
ρ0 (6.125)

⎪ 1 ∂ (β Δ θ )

⎩ b̂i (x,t) = bi (x,t) − i ∈ {1, 2, 3}
ρ0 ∂ xi
b) Constitutive equation

σ nt = C : ε = λ Tr (εε ) 1 + 2μεε (6.126)


24 The field of thermal increments Δ θ (x,t) is assumed to be known a priori and, there-
fore, independent of the mechanical response of the problem. This situation is known as the
uncoupled thermomechanical problem.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
304 C HAPTER 6. L INEAR E LASTICITY

c) Geometric equation (remains unchanged)

ε = ∇2 u (6.127)

d) Boundary condition in Γu

Γu : u = u∗ (6.128)

e) Boundary condition in Γσ

rs
ee

σ = σ nt − σ t

s gin
=⇒ σ nt · n − σ t · n = t∗ =⇒
σ · n = t∗

t d le En
(6.129)
∗ ∗
σ nt · n = t∗ + σ · n = t + β Δ θ n =⇒
t
Γσ : σ · n = t̂
nt

r
     

ba
ge ro or
eS m
β Δθ 1·n t̂∗

ci
f

ra
where t̂∗ (x,t) is a pseudo-traction vector defined by
C d P cs
b
a
i
an an n

t̂∗ = t∗ + β Δ θ n .
y ha

(6.130)
le
liv or ec

Equations (6.123) to (6.130) allow rewriting the original problem (6.121) as


M

.A



⎪ ∇ · σ nt + ρ0 b̂ = 0
m



Equilibrium
d

⎪ 1
⎪ with b̂ = b − ∇ (β Δ θ )
uu

⎪ equation

e

⎨ ρ0
X Th

er

Governing
tin

equations ⎪ Constitutive

⎪ σ nt = C : ε = λ Tr (εε ) 1 + 2μεε
⎪ equation
on


.O

⎪ (6.131)


⎩ ε = ∇S u Geometric
C

equation
©


Boundary Γu : u = u∗
conditions Γσ : σ nt · n = t̂∗ with t̂ = t + β Δ θ n

which constitutes the so-called analogous problem, a linear elastic problem that
can be solved with the methodology indicated for this type of problems in Sec-
tion 6.7 and characterized by the following actions and responses.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Thermal Analogies 305

⎫ ⎧
b̂ (x,t) ⎬  MAT HEMAT ICAL ⎨ u (x,t)

u (x,t) ⇒ MODEL : ⇒ ε (x,t)
⎭ PDEs + BCs ⎩ nt (6.132)
t̂∗ (x,t) σ (x,t)
     
not (II)
Responses = R(II) (x,t)
not
Actions = A (x,t)

Comparing the actions and responses of the original problem (6.122) with
those of the analogous problem (6.132), reveals the difference between them to

rs
be ⎡ ⎤
⎡ ⎤ ⎡ ⎤ ⎡ ⎤

ee
1
b b̂ b − b̂ ∇ (β Δ θ )
⎢ ⎥
⎢ u∗ ⎥ ⎢ 0 ⎥ ⎢ ρ0

s gin
not ⎢ u ⎥ ⎥ de f (III)

(I)
A −A (II) ⎢ ⎥ ⎢ ⎥ ⎢
A ≡ ⎣ ∗ ⎦−⎣ ∗ ⎦ = ⎣ ∗ ∗ ⎦ = ⎢ ⎥ ⎢ 0 ⎥=A (x,t)
t t̂ t − t̂ ⎥
⎣ −β Δ θ n ⎦

t d le En
Δθ 0 Δθ Δθ

r
⎡ ⎤ ⎡

ba

ge ro or
⎡ ⎤ ⎡ ⎤

eS m
ci
u u 0 0
⎢ ⎥ ⎢
f ⎥ de f (III)

ra
R(II) ≡ ⎣ ε ⎦ − ⎣ ε ⎦ = ⎢
R (I) −R ⎥=⎣
not
⎦=R (x,t)
C d P cs
⎣ 0 ⎦ 0
b
a
σ σ nt
σ −σ −β Δ θ 1
i
nt
an an n

  
y ha

−σσt
le
(6.133)
liv or ec

where (6.130) and (6.117) have been taken into account.


M

.A
m

Remark 6.13. It can be directly verified that, in (6.133), R (III) is the


uu
e

response corresponding to the system of actions A (III) in the ther-


X Th

er
tin

moelastic problem (6.121).


on

.O
C

Equation (6.133) suggests that the original problem (I) may be interpreted as
©

the sum (superposition) of two problems or states:

STATE (II) (to be solved): analogous elastic state in which the temperature
does not intervene and that can be solved by means of elastic procedures.
+
STATE (III) (trivial): trivial thermoelastic state in which the responses
R(III) (x) given in (6.133) are known without the need of any calculations.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
306 C HAPTER 6. L INEAR E LASTICITY

Once STATE (II) is computed, the solution to the original thermoelastic prob-
lem of STATE (I) is obtained as
⎧ (I) (II)
Solution to the ⎨u = u
original thermoelastic ε = ε (II)
(I)
(6.134)
problem ⎩ (I) (II)
σ = σ −β Δθ1

The procedure to solve the thermoelastic problem based on the first thermal
analogy is summarized as a superposition of states in Figure 6.13.

rs
ee
STATE ACTION RESPONSE

s gin
t d le En
⎡ ⎤
b (x,t) ⎡ ⎤

r
⎢ u∗ (x,t) ⎥ u (x,t)

ba
ge ro or
⎢ ⎥ ⎢ ⎥

eS m
⎢ ∗ ⎥ ⎣ ε (x,t) ⎦

ci
⎣ t (x,t) ⎦
f

ra
σ (x,t)
C d P cs
Δ θ (x,t)
b
a
i
an an n

(I) Thermoelastic (original)


y ha

le
liv or ec

⎡ ⎤
M

.A

1
b̂ = b − ∇ (β Δ θ )
⎢ ρ0 ⎥ ⎡ ⎤
⎢ ⎥ u (x,t)
m

⎢ ⎥
u∗ (x,t)
d

⎢ ⎥ ⎢ ⎥
⎢ ⎥ ⎣ ε (x,t) ⎦
uu

⎢ ⎥
e

⎢ ∗ ∗
t̂ = t + β Δ θ n ⎥ σ nt (x,t)
X Th

⎣ ⎦
er
tin

Δθ = 0
on

.O

(II) Elastic (analogous)


C

⎡ ⎤
+ = 1 ∇ (β Δ θ )
b ⎡ ⎤
⎢ ρ0 ⎥ u=0
⎢ ⎥
⎢ +
u ∗=0 ⎥ ⎢ ⎥
⎢ ⎥ ⎣ ε =0 ⎦
⎢ +∗ ⎥
⎣ t = −β Δ θ n ⎦ σ = −β Δ θ 1
Δ θ (x,t)

(III) Thermoelastic (trivial)

Figure 6.13: First thermal analogy.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Thermal Analogies 307

6.11.2 Second Thermal Analogy


The second thermal analogy is based on expressing the equations that constitute
the problem in terms of the thermal strains ε t defined in (6.118). Consider the
equations of the original thermoelastic problem, with the constitutive equation
in its inverse form




⎪ ∇ · σ + ρ0 b = 0 Equilibrium equation

Governing
equations ⎪ ε = C−1 : σ + α Δ θ 1 Inverse constitutive
equation


rs
⎩ ε = ∇S u Geometric equation (6.135)

ee


s gin
Boundary Γu : u = u∗
conditions Γσ : σ · n = t∗

t d le En

r
which constitute the actions (data) A (x,t) and responses (unknowns) R (x,t) of

ba
ge ro or
eS m
the problem.

ci
f

ra
C d P cs
b
a


i
b (x,t) ⎪ 
an an n



⎬ MAT HEMAT ICAL ⎨ u (x,t)
y ha

u (x,t)
⇒ MODEL : ⇒ ε (x,t)
t∗ (x,t) ⎪
le
⎪ ⎩
liv or ec

⎭ PDEs + BCs σ (x,t) (6.136)


Δ θ (x,t)
M

  
.A

  
Responses = R (I) (x,t)
not
Actions = A (I) (x,t)
not
m

d
uu
e
X Th

er
tin

Hypothesis 6.1. Assume that the coefficient of thermal expansion


on

.O

α (x) and the thermal increment Δ θ (x,t) are such that the thermal
strain field
C

ε t (x,t) = α (x) Δ θ (x,t) 1


©

is integrable (satisfies the compatibility conditions).

Consequently, there exists a thermal displacement field ut (x,t) that satisfies



⎪ 1 t

⎨ ε (x,t) = α Δ θ 1 = ∇ u = (u ⊗ ∇ + ∇ ⊗ u )
t S t t
' 2 (
1 ∂ uti ∂ u j
t (6.137)

⎪ εit j = α Δ θ δi j = + i, j ∈ {1, 2, 3}

2 ∂ x j ∂ xi

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
308 C HAPTER 6. L INEAR E LASTICITY

Remark 6.14. The solution ut (x,t) to the system of differential equa-


tions (6.137) exists if and only if the strain field ε t (x,t) satisfies the
compatibility conditions (see Chapter 3). In addition, this solution is
determined except for a rigid body motion characterized by a rota-
tion tensor Ω ∗ and a displacement vector c∗ (both constant). That is,
there exists a family of admissible solutions of the form
+ (x,t) + Ω ∗ · x + c∗
ut (x,t) = u .
   

rs
rotation translation
  

ee
rigid body motion

s gin
The rigid body motion may be chosen arbitrarily (in the form which

t d le En
is most convenient for the resolution process).

r
ba
ge ro or
eS m
ci
f
Once the thermal displacements have been defined, a decomposition of the

ra
C d P cs
total displacements into their thermal and non-thermal parts can be performed
b
a
i
as follows.
an an n
y ha

de f
unt (x,t) = u (x,t) − ut (x,t) =⇒ u = unt + ut (6.138)
le
liv or ec

To eliminate the thermal term in the equations that constitute the thermoe-
M

.A

lastic problem (6.135), the decompositions of the displacements and strains


(u = unt + ut and ε = ε nt + ε t ) is introduced in the equations of (6.135), which
m

result in
uu
e
X Th

er

a) Equilibrium equation (remains unchanged)


tin

∇ · σ + ρ0 b = 0 (6.139)
on

.O
C

b) Inverse constitutive equation


ν 1+ν
ε nt = C −1 : σ = − σ)1+
Tr (σ σ (6.140)
E E

c) Geometric equation

ε = ∇S u = ∇S (unt + ut ) = ∇S unt + ∇S ut = ∇S unt + ε t ⎬

=⇒ ε nt = ∇S unt
εt ⎭
ε = ε nt + ε t
(6.141)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Thermal Analogies 309

d) Boundary condition in Γu

u = u∗
=⇒ Γu : unt = u∗ − ut (6.142)
u = unt + ut

e) Boundary condition in Γσ (remains unchanged)

Γσ : σ · n = t∗ (6.143)

rs
Equations (6.139) to (6.143) allow rewriting the original problem (6.135) as

ee

s gin

⎪ Equilibrium

⎪ ∇ · σ + ρ0 b = 0

⎪ equation

t d le En



⎨ nt
ε = C −1 : σ =

r
Governing

ba
Constitutive

ge ro or
ν 1+ν

eS m
equations ⎪
⎪ = − Tr (σ σ)1+ σ equation

ci


⎪ E E f

ra
⎪ (6.144)

C d P cs


b
a
⎩ ε nt = ∇S unt Geometric
i
an an n

equation

y ha

Γu : u = u∗ − ut
le
Boundary
liv or ec

conditions Γσ : σ · n = t∗
M

.A
m

which constitutes the so-called analogous problem, a linear elastic problem char-
d

acterized by the following actions and responses


uu
e
X Th

er
tin

⎫  ⎧ nt
b̂ (x,t) ⎬ ⎨ u (x,t)
on

MAT HEM.
.O

u∗ (x,t) − ut (x,t) ⇒ MODEL : ⇒ ε nt (x,t)


⎭ ⎩
C

PDEs + BCs (6.145)


t̂∗ (x,t) σ (x,t)
©

     
Actions = A (II) (x,t) Responses = R (II) (x,t)
not not

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
310 C HAPTER 6. L INEAR E LASTICITY

Comparing the actions and responses of the original problem (6.136) and the
analogous problem (6.145), reveals the difference between them to be
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
b b 0
not ⎢ u ⎥ ⎢ ∗
∗ t⎥ ⎢ t ⎥
A(II) ≡ ⎢
A (I) −A ⎥ ⎢ u − u ⎥ ⎢ u ⎥ de f (III) (x,t)
⎣ t∗ ⎦ − ⎣ t∗ ⎦ = ⎣ 0 ⎦ = A
Δθ 0 Δθ
(6.146)
⎡ ⎤ ⎡ nt ⎤ ⎡ t ⎤ ⎡ t

u u u u
⎢ ⎥ de f
(I) R ≡ ⎣ ε ⎦ − ⎣ ε ⎦ = ⎣ ε ⎦ = ⎣ α Δ θ 1 ⎦ = R (III) (x,t)
(II) not
R −R nt t

rs
σ σ

ee
0 0

s gin
where equations (6.138) and (6.118) have been taken into account.

t d le En

r
ba
Remark 6.15. It can be directly verified that, in (6.146), R(III) is the

ge ro or
eS m
ci
response corresponding to the system of actions A (III) in the ther-
f

ra
C d P cs
moelastic problem (6.135).
b
a
i
an an n
y ha

Therefore, the original problem (I) can be interpreted as the sum (superposi-
le
liv or ec

tion) of two problems or states:


M

.A

STATE (II) (to be solved): analogous elastic state in which the temperature
m

does not intervene and that can be solved by means of elastic procedures.
uu
e

+
X Th

STATE (III) (trivial): trivial thermoelastic state in which the responses


er
tin

R (III) (x) given in (6.146) are known without the need of any calculations.
on

.O
C

Once STATE (II) is computed, the solution to the original thermoelastic prob-
©

lem of STATE (I) is obtained as



(I) (II)
⎨u = u +u
⎪ t
Solution to the
original thermoelastic ε (I) = ε (II) + α Δ θ 1 (6.147)
problem ⎪

σ (I) = σ (II)

where ut is known from the integration process of the thermal strain field
in (6.137). The procedure to solve the thermoelastic problem based on the sec-
ond thermal analogy is summarized as a superposition of states in Figure 6.14.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Thermal Analogies 311

STATE ACTION RESPONSE

⎡ ⎤

rs
b (x,t) ⎡ ⎤
u (x,t)

ee
⎢ u∗ (x,t) ⎥
⎢ ⎥ ⎢ ⎥
⎢ ∗ ⎥ ⎣ ε (x,t) ⎦

s gin
⎣ t (x,t) ⎦
σ (x,t)
Δ θ (x,t)

t d le En

r
(I) Thermoelastic (original)

ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
⎡ ⎤
i
b (x,t) ⎡ ⎤
an an n

⎢ u∗ − ut ⎥ unt (x,t)
y ha

⎢ ⎥ ⎢ nt ⎥
⎢ ∗ ⎥ ⎣ ε (x,t) ⎦
le
⎣ t (x,t) ⎦
liv or ec

σ (x,t)
Δθ = 0
M

.A
m

(II) Elastic (analogous)


d
uu
e
X Th

er
tin

⎡ ⎤
b=0 ⎡ ⎤
on

.O

⎢ ⎥ u = ut (x,t)
⎢ +∗ = ut
u ⎥ ⎢ ⎥
⎣ ε = α Δθ1 ⎦
C

⎢ ⎥
⎣ +t∗ = 0 ⎦
©

σ =0
Δ θ (x,t)

(III) Thermoelastic (trivial)

Figure 6.14: Second thermal analogy.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
312 C HAPTER 6. L INEAR E LASTICITY

Example 6.5 – Solve the problem of a beam fully-fixed at its ends and sub-
jected to a constant thermal increment Δ θ using the second thermal analogy.

Solution
The classic procedure followed in strength of materials to solve this problem
consists in the superposition (sum) of the following situations: 1) The struc-
ture is initially considered to be hyperstatic; 2) the right end is freed to allow
for thermal expansion, which takes place with null stresses (since it is an iso-
static structure); and 3) the displacement of the beam’s right end is recovered

rs
until it is brought again to zero.

ee
This procedure coincides exactly with the application of the second thermal

s gin
analogy in which the thermal displacement field ut is defined by the thermal
expansion of the piece with its right end freed (state III). Said expansion

t d le En
produces a displacement in the right end of value u|x= = α Δ θ  and, when
recovering the displacement at this end, the boundary condition

r
ba
ge ro or
eS m
Γu : u = u∗ − ut = −ut ,

ci
 f

ra
C d P cs
b
a
0
i
an an n

which corresponds exactly to state II of Figure 6.14, is being implicitly ap-


y ha

plied.
le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Remark 6.16. The application of the second thermal analogy essen-


tially resides in the integration of the thermal strain field ε t (x,t) to
obtain the thermal displacement field ut (x,t) (see Remark 6.14). If
the thermal strains are not integrable, the analogy cannot be applied.
Comparing its advantages and disadvantages with respect to the first
thermal analogy, it is also recommended that the integration of the
thermal strains be, in addition to possible, simple to perform.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Thermal Analogies 313

Remark 6.17. The case involving


• a homogeneous material (αα (x) = const. = α )
• a linear thermal increment (Δ θ = ax + by + cz + d)
is of particular interest. In this case, the product Δ θ α is a linear
polynomial and the thermal strains ε t = Δ θ α automatically satisfy
the compatibility conditions (6.69) (which are equations that only
contain second-order derivatives) and, therefore, the thermal strain
field is guaranteed to be integrable.

rs
ee
s gin
t d le En
Remark 6.18. In the case involving

r
• a homogeneous material (αα (x) = const. = α)

ba
ge ro or
eS m
• a constant thermal increment (Δ θ = const.)

ci
f

ra
C d P cs
the integration of the thermal strain field ε t = Δ θ α1 = const. is triv-
b
a
i
ial, resulting in
an an n
y ha

Ω ∗ · x + c∗
ut (x,t) = α Δ θ x +Ω ,
le
  
liv or ec

rigid body motion


M

.A

where the rigid body motion can be chosen arbitrarily (see Re-
m

mark 6.14). If this motion is considered to be null, the solution to


d
uu

the thermal displacement field is


e
X Th

er
tin

ut (x,t) = α Δ θ x =⇒ x + ut = x + α Δ θ x = (1 + α Δ θ ) x ,
on

.O

which means that STATE (III) in the second thermal analogy (see
C

Figure 6.15) is an homothecy, with respect to the origin of coordi-


nates, of value (1 + α Δ θ ). This homothecy is known as free thermal
©

expansion (see Figure 6.15).


The value of the thermal displacement (associated with the free ther-
mal expansion) in the boundary Γu can be trivially determined in this
case without need of formally integrating the thermal strains.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
314 C HAPTER 6. L INEAR E LASTICITY

Figure 6.15: Free thermal expansion in a homogeneous material subjected to a constant

rs
thermal increment.

ee
s gin
6.12 Superposition Principle in Linear Thermoelasticity

t d le En
Consider the linear thermoelastic problem in Figure 6.16 and its corresponding

r
governing equations

ba
ge ro or
eS m
ci
∂ 2u f

ra
C d P cs
∇ · σ + ρ0 b = ρ0 Cauchy’s equation
b
a
∂t 2
i
an an n
y ha

σ = λ Tr (εε ) 1 + 2μεε −β Δ θ 1 Constitutive equation (6.148)


  
le
liv or ec

C:ε
M

.A

1
ε = ∇S u = (u ⊗ ∇ + ∇ ⊗ u) Geometric equation
2
m

d
uu
e


X Th

er

Γu : u = u∗
tin

Boundary conditions in space (6.149)


Γσ : t∗ = σ · n
on

.O
C

Figure 6.16: Linear thermoelastic problem.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Superposition Principle in Linear Thermoelasticity 315


u (x, 0) = 0
. Initial conditions (6.150)
u (x, 0) = v0

which define the generic set of actions and responses



b̂ (x,t) ⎪⎪
⎪ ⎧
u∗ (x,t) ⎪⎪


MAT HEMAT ICAL ⎨ u (x,t)
t∗ (x,t) ⇒ MODEL : ⇒ ε (x,t)

rs

⎪ + ⎩
Δ θ (x,t) ⎪
⎪ PDEs BCs σ (x,t) (6.151)

ee

v0 (x) ⎭   

s gin
   not
Responses = R (x,t)
not
Actions = A (x,t)

t d le En

r
ba
ge ro or
eS m
ci
f

ra
Remark 6.19. The different (scalar, vector, tensor and differential)
C d P cs
b
a
operators that intervene in the governing equations of the problem
i
an an n

(6.148) to (6.150) are linear, that is, given any two scalars a and b,
y ha

le
∇ · (•) → linear =⇒ ∇ · (a x + b y) = a ∇ · x + b ∇ · y ,
liv or ec
M

.A

C : (•) → linear =⇒ C : (a x + b y) = a C : x + b C : y ,
m

∇S (•) → linear =⇒ ∇S (a x + b y) = a ∇S x + b ∇S y ,
uu
e
X Th

er

∂2 ∂ 2 (a x + b y) ∂ 2x ∂ 2y
tin

(•) → linear =⇒ = a + b .
∂t 2 ∂t 2 ∂t 2 ∂t 2
on

.O
C

Consider now two possible systems of actions A (1) and A (2) ,


⎡ (1) ⎤ ⎡ (2) ⎤
b (x,t) b (x,t)
⎢ ∗(1) ⎥ ⎢ ∗(2) ⎥
⎢ u (x,t) ⎥ ⎢ u (x,t) ⎥
⎢ ⎥ ⎢ ⎥
A (1) (x,t) ≡ ⎢ ∗(1) ⎥ (2) ⎢ ∗(2) ⎥
not not
⎢ t (x,t) ⎥ and A (x,t) ≡ ⎢ t (x,t) ⎥ , (6.152)
⎢ (1) ⎥ ⎢ (2) ⎥
⎣ Δ θ (x,t) ⎦ ⎣ Δ θ (x,t) ⎦
(1) (2)
v0 (x) v0 (x)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
316 C HAPTER 6. L INEAR E LASTICITY

and their corresponding responses, R (1) and R (2) ,


⎡ (1) ⎤ ⎡ (2) ⎤
u (x,t) u (x,t)
not ⎢ ⎥ not ⎢ ⎥
R(1) (x,t) ≡ ⎣ ε (1) (x,t) ⎦ and R (2) (x,t) ≡ ⎣ ε (2) (x,t) ⎦ . (6.153)
σ (1) (x,t) σ (2) (x,t)

Theorem 6.2. Superposition principle.


The solution (response) to the system of actions

rs
ee
A (3) = λ (1) A (1) + λ (2) A (2)

s gin
(where λ (1) and λ (2) are any two scalars) is

t d le En
R (3) = λ (1) R (1) + λ (2) R (2)

r
ba
ge ro or
eS m
ci
f
In other words, the solution to the linear thermoelastic problem

ra
C d P cs
when considering a linear combination of different systems of ac-
b
a
i
tions is the same linear combination of the individual solutions to
an an n

each of these systems of actions.


y ha

le
liv or ec
M

.A

Proof
m

Replacing the actions A (3) = λ (1) A (1) + λ (2) A (2) and the responses
d
uu

R = λ (1) R (1) + λ (2) R (2) in the equations of the problem, and taking into
(3)
e
X Th

account the linearity of the different operators (see Remark 6.19) yields
er
tin

a) Cauchy’s equation
on

.O

   
C

∇ · σ (3) + ρ0 b(3) = λ (1) ∇ · σ (1) + ρ0 b(1) +λ (2) ∇ · σ (2) + ρ0 b(2) =


©

     
∂ 2 u(1) ∂ 2 u(2)
ρ0 ρ0
∂t 2 ∂t 2
 
∂ 2 λ (1) u(1) + λ (2) u(2) ∂ 2 u(3)
= ρ0 = ρ 0
∂t 2 ∂t 2

∂ 2 u(3)
∇ · σ (3) + ρ0 b(3) = ρ0
∂t 2
(6.154)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Superposition Principle in Linear Thermoelasticity 317

b) Constitutive equation
    
σ (3) − C : ε (3) − β Δ θ (3) 1 = λ (1) σ (1) − C : ε (1) − β Δ θ (1) 1 +
  
  = 0 
λ (2) σ (2) − C : ε (2) − β Δ θ (2) 1 = 0
  
=0

σ (3) = C : ε (3) − β Δ θ (3) 1

rs
(6.155)

ee
s gin
c) Geometric equation
   

t d le En
ε (3) − ∇S u(3) = λ (1) ε (1) − ∇S u(1) +λ (2) ε (2) − ∇S u(2) = 0
     

r
ba
(6.156)

ge ro or
=0 =0

eS m
ci
f
ε (3) = ∇S u(3)

ra
C d P cs
b
a
i
an an n

d) Boundary condition in Γu
y ha

   
le
u(3) − u∗(3) = λ (1) u(1) − u∗(1) +λ (2) u(2) − u∗(2) = 0
liv or ec

     
M

.A

=0 =0 (6.157)
m

Γu : u(3) = u∗(3)
d
uu
e
X Th

er
tin

e) Boundary condition in Γσ
   
on

.O

σ (3) · n − t∗(3) = λ (1) σ (1) · n − t∗(1) +λ (2) σ (2) · n − t∗(2) = 0


     
C

=0 =0 (6.158)
©

Γσ : σ (3) · n = t∗(3)

f) Initial conditions
   
. (3) . (1) . (2)
u(3) (x, 0) − v0 = λ (1) u(1) (x, 0) − v0 +λ (2) u(2) (x, 0) − v0 = 0
     
=0 =0
. (3)
u(3) (x, 0) = v0
(6.159)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
318 C HAPTER 6. L INEAR E LASTICITY

not  T
Consequently, R (3) = λ (1) R (1) + λ (2) R (2) ≡ u(3) , ε (3) , σ (3) is the solu-
tion to the thermoelastic problem subjected to the actions: A (3) = λ (1) A (1) +
λ (2) A (2) (QED).

6.13 Hooke’s Law in terms of the Stress and Strain


“Vectors”
The symmetry of the stress and the strain tensors, σ and ε , means that only six
of its nine components in a certain Cartesian system are different. Therefore,

rs
and to “economize” in writing, only these six different components are used in

ee
engineering, and they are expressed in the form of the stress and strain “vec-

s gin
tors”. These are constructed in R6 , systematically arranging the elements of the
upper triangle of the matrix of components of the corresponding tensor in the

t d le En
following manner25 .
⎡ ⎤

r
ba
ge ro or
σx

eS m
⎢ ⎥

ci
⎡ ⎤ f ⎢ σy ⎥

ra
σx τxy τxz ⎢ ⎥
C d P cs
⎢ ⎥
b
a
not ⎢ ⎥ de f ⎢ σz ⎥
i
σ ≡ ⎣ τxy σy τyz ⎦ → {σ σ} = ⎢ ⎥ (6.160)
an an n

⎢ τxy ⎥
⎢ ⎥
y ha

τxz τyz σz ⎢τ ⎥
⎣ xz ⎦
le
liv or ec

τyz
M

.A

The same arrangement is followed in the case of the strains, with the particularity
m

that the strain vector {εε } is constructed using the angular strains γxy = 2 εxy ,
uu
e

γxz = 2 εxz and γyz = 2 εyz (see Chapter 2, Section 2.11.4).


X Th

er
tin

⎡ ⎤
⎡ ⎤ εx
on

.O

1 1 ⎢ ⎥
⎡ ⎤ ⎢ εx γxy γxz ⎥ ⎢ εy ⎥
εx εxy εxz ⎢ 2 2 ⎥ ⎢ ⎥
C

⎢ ⎥ ⎢ε ⎥
not ⎢ ⎥ not ⎢ 1 ⎥ ⎢ z⎥
©

de f
ε ≡ ⎣ εxy εy εyz ⎦ = ⎢ γxy εy 1 γyz ⎥ → {εε } = ⎢ ⎥ (6.161)
⎢2 2 ⎥ ⎢ γxy ⎥
εxz εyz εz ⎢ ⎥ ⎢ ⎥
⎣1 1 ⎦ ⎢γ ⎥
γxz γyz εz ⎣ xz ⎦
2 2 γyz

25 The notation {x} is used to denote the vector in R6 constructed from the symmetric ten-
sor x.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Hooke’s Law in terms of the Stress and Strain “Vectors” 319

Remark 6.20. An interesting property of this construction is that the


double contraction of the stress and strain tensors is transformed into
the dot product (in R6 ) of the stress and strain vectors,
σ :ε = σ } · {εε }
{σ ⇐⇒ σi j εi j = σi εi
     
second-order vectors
tensors

which can be verified by performing said operations, using the defi-

rs
nitions in (6.160) and (6.161).

ee
s gin
The inverse constitutive equation (6.113),

t d le En
ν 1+ν

r
ε =− σ)1+
Tr (σ σ + αΔ θ 1 , (6.162)

ba
ge ro or
eS m
E E

ci
f

ra
can now be rewritten in terms of the stress and strain vectors as
C d P cs
b
a
i
C−1 · {σ
{εε } = Ĉ σ } + {εε }t , (6.163)
an an n
y ha

C−1 is the inverse matrix of elastic constants,


where Ĉ
le
liv or ec

⎡ ⎤
M

.A

1 −ν −ν
⎢ 0 0 0 ⎥
⎢ E E E ⎥
m

⎢ ⎥
d

⎢ −ν 1 −ν ⎥
uu

⎢ ⎥
e

⎢ 0 0 0 ⎥
E
X Th

⎢ E E ⎥
er
tin

⎢ −ν −ν 1 ⎥
⎢ ⎥
⎢ 0 0 0 ⎥
on

not ⎢ ⎥
.O

E E E
C−1 ≡ ⎢
Ĉ ⎥ (6.164)
⎢ ⎥
C

⎢ 1 ⎥
⎢ 0 0 0 0 0 ⎥
©

⎢ G ⎥
⎢ ⎥
⎢ 1 ⎥
⎢ 0 0 0 0 0 ⎥
⎢ G ⎥
⎢ ⎥
⎣ 1 ⎦
0 0 0 0 0
G
and {εε }t is a thermal strain vector defined by means of an adequate translation
of the thermal strain tensor ε t = α Δ θ 1,

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
320 C HAPTER 6. L INEAR E LASTICITY

⎡ ⎤
α Δθ
⎡ ⎤ ⎢ ⎥
⎢α Δθ ⎥
⎢α Δθ 0 0 ⎥ ⎢



⎢ ⎥ t de f ⎢ α Δ θ ⎥
εt ≡ ⎢ ⎥
not
⎢ 0 α Δ θ 0 ⎥ → {εε } = ⎢ ⎥. (6.165)
⎣ ⎦ ⎢ 0 ⎥
⎢ ⎥
0 0 α Δθ ⎢ 0 ⎥
⎣ ⎦
0

Finally, the inversion of equation (6.163) provides Hooke’s law in terms of

rs
the stress and strain vectors,

ee
Hooke’s law  

s gin
in terms of the {σ C · {εε } − {εε }t
σ } = Ĉ (6.166)

t d le En
stress and strain vectors

r
ba
ge ro or
eS m
C is the matrix of elastic constants.
where Ĉ

ci
⎡ f ⎤

ra
C d P cs
ν ν
b
a
⎢ 1 0 0 0 ⎥
⎢ ⎥
i
1−ν 1−ν
an an n

⎢ ⎥
⎢ ν ν ⎥
y ha

⎢ 1 0 0 0 ⎥
⎢ 1−ν 1−ν ⎥
⎢ ⎥
le
⎢ ν ⎥
liv or ec

⎢ ν ⎥
⎢ 1 0 0 0 ⎥
M

⎢ 1−ν 1−ν ⎥
.A

not E (1 − ν) ⎢ ⎥ (6.167)
C≡

(1 + ν) (1 − 2ν) ⎢
⎢ 1 − 2ν


m

⎢ 0 0 0 0 0 ⎥
⎢ ⎥
d

⎢ 2 (1 − ν) ⎥
uu

⎢ ⎥
e

⎢ 1 − 2ν ⎥
X Th

⎢ 0 0 0 0 0 ⎥
er

2 (1 − ν)
tin

⎢ ⎥
⎢ ⎥
⎣ 1 − 2ν ⎦
on

0 0 0 0 0
.O

2 (1 − ν)
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 321

P ROBLEMS

Problem 6.1 – Justify whether the following statements are true or false.
a) The terms isentropic and adiabatic are equivalent when dealing with a
thermoelastic material.
b) The second thermal analogy is always applicable to linear thermoelas-

rs
tic materials.

ee
s gin
Solution

t d le En

r
a) According to the second law of thermodynamics (5.114),

ba
ge ro or
eS m
. .

ci
ρ0 θ sloc
i
= ρ0 θ s − (ρ0 r − ∇ · q) ≥ 0 .
f

ra
C d P cs
b
a
All processes are reversible in the case of a thermoelastic material and, thus, the
i
an an n

inequality becomes an equality,


y ha

. .
le
ρ0 θ sloc
i = ρ θ s − (ρ r − ∇ · q) = 0 .
liv or ec

0 0 [1]
M

.A

.
An isentropic process (entropy remains constant) is characterized by s = 0. On
m

the other hand, an adiabatoc process (variation of heat is null) satisfies


uu
e
X Th

ρ0 r − ∇ · q = 0 .
er
tin
on

Therefore, if an isentropic process is assumed, and its mathematical expression


.O

is introduced in [1], the definition of an adiabatic process is obtained,


C

.
©

ρ0 θ s − (ρ0 r − ∇ · q) = 0 =⇒ ρ0 r − ∇ · q = 0 .

=0
Conversely, if an adiabatic process is assumed, and its mathematical expression
is introduced in [1], the definition of an isentropic process is obtained,
. .
ρ0 θ s − (ρ0 r − ∇ · q) = 0 =⇒ s=0.
  
=0
In conclusion, the statement is true.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
322 C HAPTER 6. L INEAR E LASTICITY

b) The second thermal analogy is not always applicable. The condition that the
thermal strain field be integrable must be verified, that is, the thermal strain field
ε t (x,t) must satisfy the compatibility conditions (3.19),
εi j, kl + εkl, i j − εik, jl − ε jl, ik = 0 i, j, k, l ∈ {1, 2, 3} .
Given that these involve second-order derivatives of the components of the strain
tensor with respect to x, y and z, they will be automatically satisfied if α = const.
and Δ θ = const., or if α Δ θ is linear in x, y and z (which is the definition of a
linear thermoelastic material). Therefore, the statement is true.

rs
ee
s gin
Problem 6.2 – An isotropic linear elastic solid is subjected to a constant pres-
sure of value p on all of its external boundary, in addition to a thermal incre-
ment of Δ θ = θ (x, y, z) in its interior. Both actions cancel each other out such

t d le En
that no displacements are observed in the solid. Obtain the value of Δ θ in each

r
ba
point of the solid.

ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
Solution
an an n
y ha

The first thermal analogy described in Section 6.11.1 will be applied. To this
le
liv or ec

aim, the original problem I is decomposed into the sum of problems II and III as
described in Figure 6.13.
M

.A
m

P ROBLEM I
d


uu
e


⎪ b=0 ⎧

X Th

⎪ ⎪
er

⎨ t∗ = −p n ⎨u
tin

in Γσ
Actions: Responses: ε
⎪ u∗ = 0 ⎪
on

in Γu
.O


⎪ ⎩

⎩Δθ = Δθ σ
C

P ROBLEM III
This problem is solved first since its solution is trivial.

⎪ 1

⎪ bIII = β ∇ (Δ θ ) ⎧
⎪ ρ

⎨ ∗ ⎨ uIII = 0

t
Actions: III = −β Δ θ n in Γ σ Responses: ε III = 0

⎪ ∗ =0 ⎪


⎪ u in Γ u σ III = −β Δ θ 1


III
Δ θIII = Δ θ

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 323

P ROBLEM II

⎪ 1

⎪ bII = β ∇ (Δ θ )

⎪ ρ
⎨ ∗
Actions: tII = (−p + β Δ θ ) n in Γσ



⎪ u∗II = u∗ = 0 in Γu


Δ θII = 0

To solve problem II, Navier’s equation (6.62) is taken into account, together with
the fact that uII = 0.

rs
ee
(λ + μ) ∇ (∇ · uII ) + μ∇2 uII + ρbII = 0 =⇒

s gin
bII = 0 =⇒ β ∇ (Δ θ ) = 0 =⇒ Δ θ is uniform

t d le En
In addition, uII = 0 also results in

r
ba
ge ro or
eS m
ci
1 f

ra
ε II = (uII ⊗ ∇ + ∇ ⊗ uII ) = 0 ,
C d P cs
b
a
2
i
an an n

σ II = λ (∇ · uII ) 1 + μ (uII ⊗ ∇ + ∇ ⊗ uII ) = 0 .


y ha

le
liv or ec

Since the traction vector t∗II is defined in terms of the stress tensor σ II ,
M

.A

σ II · n = t∗II = (−p + β Δ θ ) n = 0 ∀n =⇒ −p + β Δ θ = 0 ,
m

d
uu

and the value of the thermal increment is finally obtained,


e
X Th

er
tin

p
Δθ = .
β
on

.O
C

Problem 6.3 – A cylindrical shell of height h, internal radius R and external


radius 2R is placed inside an infinitely rigid cylindrical cavity of height h and
radius 2R + a, with a << R. Assume the cylindrical shell is subjected to a uni-
form temperature field Δ θ .
a) Determine the value of Δ θ ∗ required for the external lateral walls of the
cylindrical shell and the rigid walls of the cavity to come into contact.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
324 C HAPTER 6. L INEAR E LASTICITY

b) Plot, indicating the most significant values, the curve δ − Δ θ , where δ


is the lengthening of the internal radius of the cylindrical shell. Deter-
mine the value of Δ θ such that this radius recovers its initial value.
c) Plot, indicating the most significant values, the curves σrr − Δ θ , σθ θ −
Δ θ and σzz − Δ θ , in points A and B.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

Hypotheses:
le
1) Young’s modulus: E
liv or ec

2) Poisson’s coefficient: ν = 0
M

.A

3) Thermal expansion coefficient: α


m

4) Isotropic linear elastic material


d
uu

5) Weights can be neglected


e
X Th

er

6) The friction between the walls is negligible


tin
on

.O
C

Solution
©

a) Two distinct phases can be identified in this problem:

First phase
The cylindrical shell has not come into contact with the rigid walls of the
cavity. The boundary condition on the lateral walls, both internal and ex-
ternal, will be null radial stress. The two cylinders will come into contact
when

ur (r = 2R) = a .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 325

Second phase
The cylindrical shell and the rigid walls of the cavity are in contact and,
therefore, the boundary condition on the external lateral wall is different
than that of the first phase. In this case, a null radial displacement will
be imposed. Nonetheless, the internal wall will retain the same boundary
condition as in the previous phase.
A positive Δ θ will reduce the internal radius since the external radius
cannot increase because it is limited by the infinitely rigid walls of the
cavity. Then, the only possibility is that the cylindrical shell continues
expanding inwards. There will be a point in which the internal radius,

rs
which had increased in the first phase, will recover its initial value.

ee
The first thermal analogy (see Section 6.11.1) and the superposition principle

s gin
(see Section 6.12) will be applied. To this aim, the original problem (problem I)

t d le En
is decomposed into the sum of problems II and II as described in Figure 6.13.
P ROBLEM III

r
ba
ge ro or
The actions in problem III, the trivial problem, are

eS m
ci
f

ra
1
C d P cs
bIII = β Δθ) .
∇ · (β
b
a
ρ
i
an an n
y ha

In this case, however, Δ θ is uniform and β is a spherical and constant tensor


le
β = β 1). Therefore,

liv or ec

bIII = 0 .
M

.A

The boundary conditions are


m

1) Prescribed displacements in Γu : uIII = 0.


uu
e

β Δ θ n = −β Δ θ n.
2) Prescribed stresses in Γσ : t = −β
X Th

er
tin

The solution to this problem is known to be


on

.O

uIII = 0
C

ε III = 0 [1]
©

σ III = −β Δ θ 1

P ROBLEM II
The actions in problem II, the analogous problem, are
1
β Δθ) .
bII = b − ∇ · (β
ρ
Here, b = 0 because the weight of the cylinder is assumed to be negligible and
the second term is zero, as seen in problem III. Therefore,
bII = 0 .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
326 C HAPTER 6. L INEAR E LASTICITY

The boundary conditions are


1) Prescribed displacements in Γu : uII = u∗ , where u∗ is the displacement
imposed in problem I.
2) Prescribed stresses in Γσ : tII = σ II · n = t∗ + β Δ θ n = t∗ + β Δ θ n,
where t∗ is the traction vector imposed in problem I.
The analogous problem will now be solved assuming an infinitesimal strains
hypothesis, since a << R and the strains are due to Δ θ , which are generally
infinitesimal. Due to cylindrical symmetry, the displacement vector u is known
to be of the form
uII (r, z) ≡ [ur (r) , 0 , uz (z)]T .
not

rs
ee
In addition, uz (z) = 0 will be imposed in all points since no information on the

s gin
top and bottom surfaces of the cylindrical shell is given. Boundary conditions
in displacements cannot be imposed for these surfaces because there is no way

t d le En
to determine the integration constants of uz that would appear if uz = 0 were
considered. Therefore, the displacement vector

r
ba
ge ro or
eS m
uII (r, z) ≡ [ur (r) , 0 , 0 ]T
not

ci
f

ra
C d P cs
b
a
is adopted. Navier’s equation (6.62) will be used to solve this problem,
i
an an n

∂ 2 uII
y ha

(λ + μ) ∇ (∇ · uII ) + μ∇2 uII + ρ0 bII = ρ0 =0.


le
∂t 2
liv or ec
M

.A

Note that the problem requires working in cylindrical coordinates and, thus, the
equation must be adapted to this system of coordinates. Given the simplifications
m

introduced into the problem, only the radial component of the equation will
d
uu

result in a non-trivial solution,


e
X Th

er
tin

∂ e 2G ∂ ωz ∂ ωθ ∂ 2 ur
(λ + 2G) − + 2G + ρbr = ρ 2 , [2]
∂r r ∂θ ∂z ∂t
on

.O
C

where br is the radial component of bII and with


©


1 ∂ ur ∂ uz
ωθ = + =0,
2 ∂z ∂r

1 1 ∂ (r uθ ) 1 ∂ ur
ωz = − =0,
2 r ∂r r ∂θ
1 ∂ (r ur ) 1 ∂ uθ ∂ uz 1 ∂ (r ur )
e= + + = .
r ∂r r ∂θ ∂z r ∂r

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 327

The values of the parameters λ , G and β that intervene in Navier’s equation


must also be determined from the known parameters (E , α , ν = 0) as follows.
λ
ν= =0 =⇒ λ =0
2 (λ + μ)
E E [3]
μ= =⇒ μ =G=
2 (1 + ν) 2
E
β= α =⇒ β = Eα
1 − 2ν

rs
ee
The problem can be considered to be a quasi-static and, taking into account
bII = 0 and the relations derived in [3], the Navier’s stokes equation [2] is re-

s gin
duced to
∂e ∂ 1 ∂

t d le En
(λ + 2G) = 0 =⇒ E (r ur ) = 0 .
∂r ∂r r ∂r

r
ba
ge ro or
Integrating this last expression leads to

eS m
ci
f

ra
1 ∂ ∂
C d P cs
(r ur ) = 2A =⇒ (r ur ) = 2Ar =⇒ r ur = Ar2 + B
b
a
r ∂r ∂r
i
[4]
an an n

! "T
y ha

B not B
=⇒ ur = Ar + =⇒ uII (r) ≡ Ar + , 0 , 0 ,
le
r r
liv or ec
M

.A

where A and B are the integration constants. The strain tensor corresponding
to this displacement vector is easily obtained by means of the geometric equa-
m

tion (6.3),
uu

⎡ ⎤
e
X Th

B
er

A− 2
tin

0 0
⎢ r ⎥
not ⎢ ⎥
ε II (r) ≡ ⎢ 0 A + B 0 ⎥ . [5]
on

.O

⎣ r2 ⎦
C

0 0 0
©

Finally, the stress tensor is obtained through the constitutive equation of an


isotropic linear elastic material (6.20), particularized with the expressions in [3],

σ II = λ (Tr (εε II )) 1 + 2μεε II =⇒ σ II = Eεε II . [6]

First phase
The integration constants A and B must be determined by means of the
boundary conditions. Stresses can be imposed in both lateral walls of the
cylindrical shell as follows.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
328 C HAPTER 6. L INEAR E LASTICITY

B OUNDARY CONDITION AT r = 2R
If r = 2R and according to the boundary conditions in Γσ of the analogous
problem,
tII = σ II · n = t∗ + β Δ θ n = t∗ + β Δ θ n .
Here, the following is known:
n = [1 , 0 , 0 ]T : outward unit normal vector.
t∗ = 0, since, for this phase, problem I has no loading on the lateral walls.
σ II is given by [5] and [6].

rs
Therefore, the boundary condition is reduced to

ee
σrr (r = 2R) = β Δ θ ,

s gin
which, replacing the value of the radial stress from [6] and, consider-

t d le En
ing [3], results in

r
ba
ge ro or
eS m
B
A− = α Δθ .

ci
[7]
4R2 f

ra
C d P cs
b
a
i
an an n

B OUNDARY CONDITION AT r = R
y ha

If r = R and according to the boundary conditions in Γσ of the analogous


le
liv or ec

problem,
M

.A

tII = σ II · n = t∗ + β Δ θ n = t∗ + β Δ θ n .
m

Here, the following is known:


d
uu

n = [−1 , 0 , 0 ]T : outward unit normal vector.


e
X Th

er

t∗ = 0, since, for this phase, problem I has no loading on the lateral walls.
tin

σ II is given by [5] and [6].


on

.O

Therefore, the boundary condition is reduced to


C

σrr (r = R) = β Δ θ ,
which, replacing the value of the radial stress from [6] and, consider-
ing [3], results in

B
A− = α Δθ . [8]
R2
From [7] and [8], the values

A = α Δθ and B=0 [9]

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 329

are obtained. Now, replacing [9] in [4], [5] and [6] results in the displace-
ments, strains and stresses of the analogous problem.
not  T
uII ≡ α Δ θ r , 0 , 0
⎡ ⎤
α Δθ 0 0
not ⎢ ⎥
ε II ≡ ⎣ 0 α Δ θ 0 ⎦
0 0 0 [10]
⎡ ⎤
Eα Δ θ

rs
0 0
not ⎢ ⎥

ee
σ II ≡ ⎣ 0 Eα Δ θ 0 ⎦

s gin
0 0 0

t d le En
Taking into account the superposition principle (see Section 6.12), and

r
expressions [1], [3] and [10], the original problem is solved for the first

ba
ge ro or
eS m
phase.
not  T

ci
u ≡ α Δθr, 0, 0 f

ra
C d P cs
b
a
⎡ ⎤
i
α Δθ 0 0
an an n

not ⎢ ⎥
y ha

ε ≡ ⎣ 0 α Δθ 0 ⎦
le
liv or ec

0 0 0 [11]
⎡ ⎤
M

.A

0 0 0
not ⎢ ⎥
m

σ ≡⎣ 0 0 0 ⎦
d
uu
e

0 0 −Eα Δ θ
X Th

er
tin

To obtain the value of Δ θ ∗ for which the external lateral walls of the cylindrical
on

.O

shell and the rigid walls of the cavity come into contact, it is enough to impose
C

that
ur (r = 2R) = a =⇒ α Δ θ ∗ 2R = a .
©

Then, the temperature field required for the external lateral walls of the cylindri-
cal shell and the rigid walls of the cavity to come into contact is
a
Δθ∗ = . [12]
2αR

b) First, the value Δ θ ∗∗ for which the internal radius recovers its initial position
will be determined. To this aim, the same geometry as in the initial problem will
be used, but now there will exist contact between the cylindrical shell and the

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
330 C HAPTER 6. L INEAR E LASTICITY

rigid walls of the cavity, which corresponds to the second phase defined in the
previous section. So, a new problem must be solved, with the same geometry as
before but considering different boundary conditions.

Second phase
The first phase will be obviated in this section, but one must bear in mind
that the solid now starts from a state that results from the previous phase,
that is, it has already suffered certain displacements, strains, stresses and
thermal increments. The variable Δ θ will be used.

rs
As before, the first thermal analogy will be applied. Problem III remains
unchanged and, thus, so does its result [1]. Therefore, problem II must be

ee
solved with the same expressions [4], [5] and [6]. The integration con-

s gin
stants A and B must be determined by means of the boundary conditions.
Stresses can be imposed on the internal lateral wall of the cylindrical shell

t d le En
and displacements, on its external lateral wall.

r
ba
ge ro or
B OUNDARY CONDITION AT r = 2R

eS m
ci
f
If r = 2R and according to the boundary conditions in Γu of the analogous

ra
C d P cs
b
a
problem,
i
ur (r = 2R) = 0 .
an an n
y ha

Therefore, the following condition is obtained,


le
liv or ec

B
A 2R + =0 . [13]
M

.A

2R
m

d
uu

B OUNDARY CONDITION AT r = R
e
X Th

er
tin

If r = R and according to the boundary conditions in Γσ of the analogous


problem,
on

.O

tII = σ II · n = t∗ + β Δ θ n = t∗ + β Δ θ n .
C

Here, the following is known:


©

n ≡ [−1 , 0 , 0 ]T : outward unit normal vector.


not

t∗ = 0, since, for this phase, problem I has no loading on the lateral walls.
σ II is given by [5] and [6].
Therefore, the boundary condition is reduced to
σrr (r = R) = β Δ θ ,
which, replacing the value of the radial stress from [6], and consider-
ing [3], results in
B
A− 2 = α Δθ . [14]
R

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 331

From [13] and [14], the values

1 4
A = α Δθ and B = − α Δ θ R2 [15]
5 5
are obtained. Introducing now [15] in [4], [5] and [6] results in the dis-
placements, strains and stresses of the analogous problem.
! "T
1
not 4R2
uII ≡ α Δθ r − , 0, 0
5 r

rs
⎡ ⎤

ee
1 4R2
⎢ 5 α Δ θ 1 + r2 0 0 ⎥

s gin
⎢ ⎥
not ⎢ 1 4R2 ⎥
ε II ≡ ⎢ 0 α Δθ 1− 2 0 ⎥
⎢ ⎥

t d le En
⎣ 5 r ⎦

r
[16]

ba
0 0 0

ge ro or
eS m
ci
⎡ f ⎤

ra
C d P cs
1 4R2
b
a
⎢ 5 Eα Δ θ 1 + r2 0 0 ⎥
i
⎢ ⎥
an an n


not ⎢ 4R2 ⎥
y ha

σ II ≡ ⎢ 1 ⎥
⎢ 0 Eα Δ θ 1 − 0 ⎥
le
⎣ 5 r2 ⎦
liv or ec

0 0 0
M

.A

Taking into account the superposition principle (see Section 6.12), and
m

expressions [1], [3] and [16], the original problem is solved for the second
uu
e

phase. ! "T
X Th

er

4R2
tin

not 1
u ≡ α Δθ r − , 0, 0
5 r
on

.O

⎡ ⎤
4R2
C

1
⎢5 α Δ θ 1 + 0 0 ⎥ [17a]
©

⎢ r2 ⎥
not ⎢ 1 4R 2 ⎥
ε ≡⎢ 0 α Δθ 1− 2 0 ⎥
⎢ 5 r ⎥
⎣ ⎦
0 0 0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
332 C HAPTER 6. L INEAR E LASTICITY

⎡ ⎤
4 R2
⎢ 5 Eα Δ θ −1 + r2 0 0 ⎥
⎢ ⎥
not ⎢ 4 R 2 ⎥
σ ≡⎢ 0 Eα Δ θ −1 − 2 0 ⎥ [17b]
⎢ 5 r ⎥
⎣ ⎦
0 0 −Eα Δ θ
Note that, up to this point, the second phase has been solved assuming
an initial neutral state. In reality, this phase starts from the final state of
the first phase, which has the displacements, strains, stresses and thermal

rs
increments corresponding to Δ θ = Δ θ ∗ ,

ee
uinitial = u f irst phase (Δ θ = Δ θ ∗ ) ,

s gin
ε initial = ε f irst phase (Δ θ = Δ θ ∗ ) , [18]

t d le En
σ initial = σ f irst phase (Δ θ = Δ θ ∗ ) .

r
ba
ge ro or
In fact, the variable Δ θ in [17] is not a total thermal increment but the

eS m
ci
difference in temperature at the moment corresponding to Δ θ ∗ , that is,
f

ra
C d P cs
b
a
Δθ = Δθ −Δθ∗ .
i
[19]
an an n
y ha

Then, considering [17], [18] and [19], the actual displacements, strains
le
liv or ec

and stresses during the second phase are obtained,


 
M

.A

usecond phase = uinitial + u Δ θ ,


 
ε second phase = ε initial + ε Δ θ ,
m

[20]
 
d
uu

σ second phase = σ initial + σ Δ θ .


e
X Th

er
tin

Therefore, to determine Δ θ ∗∗ , it is enough to impose that the displacement,


according to the first phase, of the internal radius be equal but of opposite sign
on

.O

to that of the second phase. In this way, the total displacement will be null.
C

First phase: displacement for r = R and Δ θ = Δ θ ∗ . From [11] and [12],


©

a
δ1 = ur (r = R, Δ θ = Δ θ ∗ ) = α Δ θ ∗ R = . [21]
2

Second phase: displacement for r = R and Δ θ = Δ θ ∗∗ . From [17],


  3
δ2 = ur r = R, Δ θ = Δ θ ∗∗ = − α Δ θ ∗∗ R . [22]
5
Then,
5a
δ1 = −δ2 =⇒ Δ θ ∗∗ = . [23]
6αR

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 333

Finally, from [19] the total thermal increment is obtained,

5a a 4a
Δ θ ∗∗ = Δ θ ∗∗ + Δ θ ∗ = + =⇒ Δ θ ∗∗ =
6αR 2αR 3αR

Now, the curve δ −Δ θ can be plotted, where δ is the displacement of the internal
radius of the cylindrical shell.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A

c) Expressions [11] and [17] must be used to plot the curves σrr − Δ θ ,
m

σθ θ − Δ θ and σzz − Δ θ for points B (r = R) and A (r = 2R).


d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
334 C HAPTER 6. L INEAR E LASTICITY

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 335

E XERCISES

6.1 – A cylinder composed of an isotropic linear elastic material stands on a


rigid base. At a very small distance “a” (a  h) of its top face there is another
rigid surface. A uniform pressure p acts on all the lateral surface of the cylinder.
Plot, indicating the most significant values, the following curves:
a) Curve p − δ , where δ is the shortening of the radius of the cylinder, R.

rs
b) Curve p − σA , where σA is the stress normal to the bottom contact sur-

ee
face at point A.

s gin
t d le En

r
ba
ge ro or
eS m
Additional hypotheses:

ci
f

ra
C d P cs
1) Weights can be neglected.
b
a
2) Lamé’s constants: λ = μ
i
an an n
y ha

3) The problem is assumed


le
to be quasi-static.
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

6.2 – The solid sphere A with external radius R1 and the solid spherical B, with
external radius R2 are composed of the same material. The external surface of
A and the internal surface of B are separated by a very small distance “a”
(a  R1 and a  R2 ).
a) Determine what value of the uniform normal pressure p shown in the
figure is required for the two surfaces to be in contact.
b) Plot, indicating the most significant values, the curve p − δ , where δ is
the shortening of R2 .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
336 C HAPTER 6. L INEAR E LASTICITY

Additional hypotheses:
1) Young’s modulus: E
2) Lamé’s constants: λ = μ
3) R1 = R
4) R2 = 2 R

rs
ee
s gin
t d le En

r
6.3 – Two solid cylinders composed of different elastic materials are vertically

ba
ge ro or
eS m
superimposed and confined between two infinitely rigid walls. The cylinders are

ci
f
subjected to the external pressures p and α p (p > 0, α > 0) as shown in the

ra
C d P cs
figure.
b
a
i
an an n

a) Determine the displacement field of the two cylinders in terms of the


y ha

integration constants (justify the assumptions used).


le
liv or ec

b) Indicate the boundary conditions that need to be applied for the different
M

boundaries of the problem.


.A

c) Assuming a constant value α such that the contact surface between the
m

two cylinders does not have a vertical displacement, calculate the inte-
d
uu

gration constants and the value of α.


e
X Th

er
tin
on

.O
C

Additional hypotheses:
©

1) Top cylinder: λ1 = μ1
2) Bottom cylinder: λ2 = μ2
3) The friction between the cylin-
ders and between the cylinders
and the walls is assumed to be
null.
4) Weights can be neglected.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 337

6.4 – A cylinder with radius Ri is placed in the interior of a cylindrical shell


with internal radius Ri + 2e and external radius Re . There is an elastic gasket
between the cylinder and the cylindrical shell which has an internal radius Ri
and a thickness “e”. The cylindrical shell is subjected to an external pressure p.
a) Determine the displacement, strain and stress fields of the cylinder and
the cylindrical shell.
b) Plot the curves Ur − p, where Ur is the radial displacement, and σrr − p,
where σrr is the radial stress at points A, B and C of the figure.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n

Data:
y ha

Ri = 1
le
liv or ec

Re = 2
M

.A

ν =0
m

E (Young’s modulus)
d
uu
e
X Th

er
tin

Additional hypotheses:
1) The constitutive law of the elastic gasket is p∗ = K δ ∗ , where p∗ is the pres-
on

.O

sure acting on the gasket, δ ∗ is the shortening of its thickness and K is its
C

elastic modulus.
©

2) e  Ri
3) A plane strain behavior in an infinitesimal strain framework may be as-
sumed.

6.5 – The figure below schematizes the layout of a railway rail composed of
straight rails of length “L”, separated by an elastic gasket with elastic modu-
lus K. Due to symmetry and construction considerations, it can be assumed that
the section x = 0 suffers no longitudinal displacements and the inferior part of
the rail suffers no vertical displacements. A constant thermal increment Δ θ is
imposed in all points of the rail.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
338 C HAPTER 6. L INEAR E LASTICITY

a) Obtain the displacement, strain and stress fields in terms of the corre-
sponding integration constants.
b) Indicate the boundary conditions that must be applied to determine the
integration constants.
c) Determine the integration constants and obtain the corresponding dis-
placement, strain and stress fields.
d) Particularize these results for the cases K = 0 (open junction) and
K → ∞ (continuous rail).

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
Additional hypotheses:
f

ra
C d P cs
Assume the displacements are of the form u = [u (x) , v (y) , w (z) ]T .
b
1)

a
i
2) Linear elastic material
an an n
y ha

3) λ =μ
le
4) The weight of the rail can be neglected.
liv or ec
M

.A

6.6 – A solid cylinder with radius R and height h is placed between two in-
m

finitely rigid walls, fitting perfectly between them without producing any stress.
d
uu

A thermal increment Δ θ > 0 is applied on the cylinder. Determine:


e
X Th

er
tin

a) The displacement field in terms of the corresponding integration con-


stants.
on

.O

b) The integration constants.


C

c) The stress state. Plot its variation along the radius.


©

Additional hypotheses:
1) Material properties: λ = μ and
α = α (r) = α0 + α1 r
2) The friction between the cylinder
and the walls is negligible.
3) Weights can be neglected.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
CH.7. PLANE LINEAR
ELASTICITY
Multimedia Course on Continuum Mechanics
Overview
 Plane Linear Elasticity Theory Lecture 1
 Plane Stress
 Simplifying Hypothesis
 Strain Field Lecture 2
 Constitutive Equation
 Displacement Field
 The Linear Elastic Problem in Plane Stress Lecture 3
 Examples
 Plane Strain
 Simplifying Hypothesis
 Strain Field
 Constitutive Equation Lecture 4
 Stress Field
 The Linear Elastic Problem in Plane Stress
 Examples

2
Overview (cont’d)
 The Plane Linear Elastic Problem
Lecture 5
 Governing Equations
 Representative Curves
 Isostatics or stress trajectories
 Isoclines
 Isobars Lecture 6
 Maximum shear lines

3
7.1 Plane Linear Elasticity Theory
Ch.7. Plane Linear Elasticity

4
Plane Linear Elasticity
 For some problems, one of the principal directions is known
a priori:
 Due to particular geometries, loading and boundary conditions
involved.
 The elastic problem can be solved independently for this direction.
 Setting the known direction as z, the elastic problem analysis is
reduced to the x-y plane PLANE ELASTICITY

 There are two main classes of plane linear elastic problems:


 Plane stress
REMARK
 Plane strain The isothermal case will not be studied here for the
sake of simplicity. Generalization of the results
obtained to thermo-elasticity is straight-forward.

7
7.2 Plane Stress
Ch.7. Plane Linear Elasticity

8
Hypothesis on the Stress Tensor
 Simplifying hypothesis of a plane stress linear elastic problem:
1. Only stresses “contained in the x-y plane” are not null

σ x τ xy 0
 
[σ ]xyz ≡ τ xy σ y 0
0 0 0 

2. The stress are independent of the z direction.

σ x = σ x ( x, y , t )
σ y = σ y ( x, y , t ) REMARK
τ xy = τ xy ( x, y, t ) The name “plane stress” arises
from the fact that all (not null)
stress are contained in the x-y
plane.

9
Geometry and Actions in Plane Stress
 These hypothesis are valid when:
 The thickness is much smaller than the typical dimension associated to the
plane of analysis: e << L
 The actions b ( x,t ) , u* ( x,t ) and t* ( x,t ) are contained in the plane of
analysis (in-plane actions) and independent of the third dimension, z.
 t* ( x,t ) is only non-zero on the
contour of the body’s thickness:

10
Strain Field in Plane Stress
 The strain field is obtained from the inverse Hooke’s Law:
2(1 + ν )
(σ x − νσ y )
1
εx = γ xy =
2ε xy = τ xy
ν 1 +ν E E
− Tr ( σ ) 1 +
ε= σ
E E
σ z =0
1
εy =
E
(σ y − νσ x ) γ xz =
2ε xz =
0
τ xz =0
ν
τ yz =0 − (σ x + σ y )
εz = γ yz =
2ε yz =
0
E

 As σ x = σ x ( x, y , t )
ε = ε ( x, y , t )
σ y = σ y ( x, y , t )

 And the strain tensor for plane stress is:


 1 
 εx γ xy 0
2
ν

ε ( x, y, t ) ≡  γ xy
1
εy 0

with ε z =

1 −ν
( εx +εy )
2 
 
 0 0 εz 
 
11
Constitutive equation in Plane Stress
 Operating on the result yields:
E
2(1 + ν ) =σx ε x + νε y 
1
εx =
E
(σ x − νσ y ) γ xy 2ε xy =
=
E
τ xy (1 − ν )
2

(σ y − νσ x )
1 E
εy = γ xz =
2ε xz = =σy ε y + νε x 
(1 − ν )
0 2
E
ν
− (σ x + σ y )
εz = γ yz =
2ε yz =
0 τ xy =
E
γ xz
E plane
2 (1 + ν )
=C stress

σ x  1 ν 0  ε x 
  E ν 1   
σ y  =  0  ε y 
τ  1 −ν
2
 1 −ν  γ 
 xy   xy 
 0 0
2 
= {σ} = {ε}

Constitutive equation
{σ} C stress ⋅ {ε}
plane

in plane stress =
(Voigt’s notation)
12
Displacement Field in Plane Stress
 The displacement field is obtained from the geometric equations,
ε ( x, t ) = ∇ S u ( x, t ) . These are split into:
 Those which do not affect the displacement u z :
∂u x
ε x ( x, y , t ) =
∂x
∂u y Integration u x = u x ( x, y , t )
ε y ( x, y , t ) =
∂y in . u y = u y ( x, y , t )
∂u x ∂u y
γ xy ( x, y=
, t ) 2ε= +
∂y ∂x
xy

 Those in which u z appears:


ν ∂uz
ε z ( x, y , t ) =

1 −ν
( ε x + ε y ) ( x, y , t ) =
∂z
⇒ uz ( x, y , z, t )

∂u x ( x, y ) ∂uz ∂uz 
γ xz ( x, y , t ) = 2ε xz =
∂z  ∂x
+ =
∂x
= 0 Contradiction !!!
 
=0 
 ⇒ uz ( z, t )
∂u y ( x, y ) ∂uz ∂uz
γ yz ( x, y , t ) = 2ε yz = + = = 0
 ∂ z  ∂y ∂y 
=0 
13
The Linear Elastic Problem in
Plane Stress
 The problem can be reduced to the two dimensions of the plane of
analysis.
 The unknowns are:
ε x  σ x 
u x 
u ( x, y , t ) ≡   {ε}( x, y, t ) ≡ ε y  {σ}( x, y, t ) ≡ σ y 
u y  γ  τ 
 xy   xy 
 The additional unknowns (with respect to the general problem) are either null,
or independently obtained, or irrelevant:

σ=
z τ=
xz τ=
yz γ=
xz γ=
yz 0 REMARK
This is an ideal elastic problem because it
ν
εz = −
1 −ν
(ε x + ε y ) cannot be exactly reproduced as a particular
case of the 3D elastic problem. There is no
u z ( x, y , z , t ) does not appear guarantee that the solution to u x ( x, y, t ) and
in the problem
u y ( x, y, t ) will allow obtaining the solution to
u z ( x, y, z , t ) for the additional geometric eqns.

14
Examples of Plane Stress Analysis
 3D problems which are typically assimilated to a plane stress state are
characterized by:
 One of the body’s dimensions is significantly smaller than the other two.
 The actions are contained in the plane formed by the two “large” dimensions.

Slab loaded on
Deep beam
the mean plane

15
7.3 Plane Strain
Ch.7. Plane Linear Elasticity

16
Hypothesis on the Displacement Field
 Simplifying hypothesis of a plane strain linear elastic problem:
1. The displacement field is

u x 
 
u = u y 
0
 

2. The displacement variables associated to the x-y plane are


independent of the z direction.

u x = u x ( x, y , t )
u y = u y ( x, y , t )

17
Geometry and Actions in Plane Strain
 These hypothesis are valid when:
 The body being studied is generated by moving the plane of analysis
along a generational line.
 The actions b ( x,t ) , u* ( x,t ) and t* ( x,t ) are contained in the plane
of analysis and independent of the third dimension, z.
 In the central section, considered as the “analysis section” the
following holds (approximately) true:

uz = 0
∂u x
=0
∂z
∂u y
=0
∂z

18
uz = 0
∂u x

Strain Field in Plane Strain


=0
∂z
∂u y
=0
∂z

 The strain field is obtained from the geometric equations:


∂u x ( x, y , t ) ∂uz
ε x ( x, y=
,t) εz
= = 0
∂x ∂z
∂u y ( x, y , t ) ∂u x ( x, y , t ) ∂uz
ε y ( x=
, y, t ) γ xz
= =
+ 0
∂y ∂z ∂x
∂u x ( x, y , t ) ∂u y ( x, y , t ) ∂u y ( x, y , t ) ∂u
γ xy ( x=
, y, t ) + γ yz
= =
+ z 0
∂y ∂x ∂z ∂y
 And the strain tensor for plane strain is:
 1 
 εx 2
γ xy 0
 
REMARK
ε ( x, y, t ) ≡  γ xy 0
1
εy
2  The name “plane strain” arises
 
 0 0 0 from the fact that all strain is
  contained in the x-y plane.

19
Stress Field in Plane Strain
 Introducing the strain tensor into Hooke’s Law
= (σ λTr ( ε ) 1 + 2Gε ) and
operating on the result yields:
σ x = λ (ε x + ε y ) + 2Gε x τ xy = G γ xy
σ y = λ (ε x + ε y ) + 2Gε y τ xz = G γ xz = 0
= ( λ + 2G ) ε y + λε x
σ z = λ (ε x + ε y ) = v (σ x + σ y ) τ yz = G γ yz = 0

ε x = ε x ( x, y , t )
 As ε y = ε y ( x, y , t )
σ = σ ( x, y , t )
ε z = ε z ( x, y , t )
γ xy = γ xy ( x, y , t )

σ x τ xy 0 
And the stress tensor
σ z ν (σ x + σ y )

σ ( x, y, t ) ≡ τ xy σ y 0  with=
for plane strain is:  0 0 σ z 

20
Constitutive equation in Plane Strain
 Introducing the values of the strain tensor into the constitutive equation
and operating on the result yields:
E (1 −ν )  ν 
σ x =( λ + 2G ) ε x + λε y = ε + εy
(1 +ν )(1 − 2ν )  1 −ν 
x

σ λ Tr ( ε ) 1 + 2 µ ε σ y =( λ + 2G ) ε y + λε x =
E (1 −ν ) 
ε +
ν
ε

(1 +ν )(1 − 2ν )  1 −ν 
y x

E
τ xy G=
= γ xy γ xy
2 (1 +ν )
plane
=C strain

 ν 
 1 0 
1 −ν
{σ} C strain ⋅ {ε}
plane
σ x    ε x  =
  E (1 − ν )  ν   
σ y  =  1 0  ε y 
τ  (1 +ν )(1 − 2ν ) 1 −ν  γ  Constitutive equation
 xy   0 1 − 2ν   xy  in plane strain
= {σ} 

0
2 (1 −ν )  = {ε} (Voigt’s notation)

21
The Lineal Elastic Problem in
Plane Strain (summary)
 The problem can be reduced to the two dimensions of the plane of
analysis.
 The unknowns are:
ε x  σ x 
u x 
u ( x, y , t ) ≡   {ε}( x, y, t ) ≡ ε y  {σ}( x, y, t ) ≡ σ y 
u y  γ  τ 
 xy   xy 

 The additional unknowns (with respect to the general problem) are either null
or obtained from the unknowns of the problem:
uz = 0
ε=
z γ=
xz γ=
yz τ=
xz τ=
yz 0
σ z ν (σ x + σ y )
=

22
Examples of Plane Strain Analysis
 3D problems which are typically assimilated to a plane strain state are
characterized by:
 The body is generated by translating a generational section with actions
contained in its plane along a line perpendicular to this plane.
 The plane strain hypothesis ( ε=
z γ=
xz γ=
yz 0 ) must be justifiable. This typically
occurs when:
1. One of the body’s dimensions is significantly larger than the other two.
Any section not close to the extremes can be considered a symmetry
plane and satisfies:
uz = 0
∂u x u x 
=0  
∂z u = u y 
∂u y 0
=0  
∂z
2. The displacement in z is blocked at the extreme sections.

23
Examples of Plane Strain Analysis
 3D problems which are typically assimilated to a plane strain state are:

Pressure pipe Continuous brake shoe

Solid with blocked z


Tunnel
displacements at the ends

24
7.4 The Plane Linear Elastic Problem
Ch.7. Plane Linear Elasticity

25
Plane problem
 A lineal elastic solid is subjected to body forces and prescribed
traction and displacement
Actions:

 t *
( x, y, t ) 
On Γσ : t = *
* x

 y
t ( x , y , t ) 

x ( x, y , t ) 

 u *

On Γu : u* =  * 
 y (
u x , y , t )
bx ( x, y, t ) 
On Ω: b=
( ) 
 y
b x , y , t 

 The Plane Linear Elastic problem is the set of equations that


allow obtaining the evolution through time of the corresponding
displacements u ( x, y, t ), strains ε ( x, y, t ) and stresses σ ( x, y, t ) .
26
Governing Equations
 The Plane Linear Elastic Problem is governed by the equations:
1. Cauchy’s Equation of Motion.
Linear Momentum Balance Equation.
∂ 2 u ( x, t )
∇ ⋅ σ ( x, t ) + ρ 0 b ( x, t ) =
ρ0
∂ t2
2D

∂σ x ∂τ xy ∂τ xz ∂ 2u x
+ + + ρ bx =
ρ 2
∂x ∂y ∂z ∂t
∂τ xy ∂σ y ∂τ yz ∂ 2u y
+ + + ρ by =
ρ 2
∂x ∂y ∂z ∂t
∂τ xz ∂τ yz ∂σ z ∂ 2u z
+ + + ρ bz =
ρ 2
∂x ∂y ∂z ∂t

27
Governing Equations
 The Plane Linear Elastic Problem is governed by the equations:
2. Constitutive Equation (Voigt’s notation).
Isotropic Linear Elastic Constitutive Equation.
σ ( x, t ) = C : ε
2D

{σ}= C ⋅ {ε}
σ x  εx  1 ν 0 
With {σ} ≡ σ y  , {ε} =  ε y  and C=
E 
2 
ν 1 0


τ  γ  1 −ν
 xy   xy   0 0 (1 −ν ) 2 
E
E=
PLANE E=E PLANE 1 −ν 2
STRESS ν =ν STRAIN ν
ν =
(1 −ν )
28
Governing Equations
 The Plane Linear Elastic Problem is governed by the equations:
3. Geometrical Equation.
Kinematic Compatibility.
1
ε ( x, t )= ∇ S u ( x, t )= (u ⊗ ∇ + ∇ ⊗ u )
2

2D
This is a PDE system of
∂u
εx = x 8 eqns -8 unknowns:
∂x
∂u u ( x,t ) 2 unknowns
εy = y
∂y ε ( x,t ) 3 unknowns
γ=
∂u x ∂u y
+
σ ( x,t ) 3 unknowns
∂y ∂x
xy
Which must be solved in
the  2 ×  + space.

29
Boundary Conditions
 Boundary conditions in space
 Affect the spatial arguments of the unknowns
 Are applied on the contour Γ of the solid,
which is divided into:
 Prescribed displacements on Γ u :

x ( x, y , t ) 

 u *
= u *

u = *
* x

 y
u = u *
y ( x , y , t ) 

 Prescribed stresses on Γσ : nx 


n= 
n y 

 t *
= t *
( x, y, t )  t*= σ ⋅ n with
t = * *
* x x

 y y (
t = t x , y , t )  σ x τ xy 
σ≡ 
τ xy σ y 

30
Boundary Conditions
 INTIAL CONDITIONS (boundary conditions in time)
 Affect the time argument of the unknowns.
 Generally, they are the known values at t = 0 :
 Initial displacements:
ux 
u ( x, =
y ,0 ) =  0
u
 y
 Initial velocity:

∂u ( x, y , t )  u x   v x 
≡ u ( x, y ,0 ) =
   =  = v 0 ( x, y )
∂t t =0
u
 y  y v

31
Unknowns
 The 8 unknowns to be solved in the problem are:
 1 
u x   εx γ xy
2  σ x τ xy 
u ( x, y , t ) =   ε ( x, y , t ) ≡   σ ( x, y , t ) ≡  
u y  1 γ τ σ
εy   xy y 
 2 xy 

 Once these are obtained, the following are calculated explicitly:


ν
=
PLANE STRESS εz
1 −ν
( ε x +εy )

σz ν σx +σ y
PLANE STRAIN = ( )

32
7.5 Representative Curves
Ch.7. Plane Linear Elasticity

33
Introduction
 Traditionally, plane stress states where graphically represented
with the aid of the following contour lines:
 Isostatics or stress trajectories
 Isoclines
 Isobars
 Maximum shear lines
 Others: isochromatics, isopatchs, etc.

34
Isostatics or Stress Trajectories
 System of curves which are tangent to the principal axes of stress
at each material point .
 They are the envelopes of the principal stress vector fields.
 There will exist two (orthogonal) families of curves at each point:
 Isostatics σ 1 , tangents to the largest principal stress.
 Isostatics σ 2 , tangents to the smallest principal stress.

REMARK
The principal stresses are orthogonal
to each other, therefore, so will the two
families of isostatics orthogonal to
each other.

35
Singular and Neutral Points
 Singular point: characterized by the stress state
σx =σy
τ xy = 0
 Neutral point: characterized by the stress state
σ=
x σ=
y τ=
xy 0

Mohr’s Circle of
REMARK
a singular point
In a singular point, all directions
are principal directions. Thus, in
singular points isostatics tend to
Mohr’s Circle of loose their regularity and can
a neutral point abruptly change direction.

36
Differential Equation of the Isostatics
 Consider the general equation of an isostatic curve: y = f ( x )

2τ xy 2 tg α
=
tg ( 2α ) =
σ x − σ y 1 − tg 2 α
dy
α
tg= = y′
dx

2τ xy 2 y′ σ x −σ y
= ( y′ ) + y′ − 1 =0
2

σ x −σ y 1 − ( y′ ) τ xy
2

φ ( x, y ) Known this function,


 Solving the 2nd order eq.: the eq. can be integrated
to obtain a family of
Differential equation y ' =

( σ x −σ y )  σ x −σ y 
± 
2

+ 1 curves of the type:


 2τ 

of the isostatics 2τ xy  xy  = y f ( x) + C

37
Isoclines
 Locus of the points along which the principal stresses are in the
same direction.
 The principal stress vectors in all points of an isocline are parallel to
each other, forming a constant angle θ with the x-axis.

 These curves can be directly found using photoelasticity methods.

38
Equation of the Isoclines
 To obtain the general equation of an isocline with angle θ , the
principal stress σ 1 must form an angle α = θ with the x-axis:
ϕ ( x, y )
2τ xy For each value of θ , the equation of
Algebraic equation
tg ( 2θ ) =
of the isoclines σ x −σ y the family of isoclines parameterized
in function of θ is obtained:
y = f ( x, θ )

REMARK
Once the family of isoclines is
known, the principal stress
directions in any point of the
medium can be obtained and,
thus, the isostatics calculated.

39
Maximum shear lines
 Envelopes of the maximum shear stress (in modulus) vector fields.
 They are the curves on which the shear stress modulus is a maximum.
 Two planes of maximum shear stress correspond to each material
point, τ max and τ min .
 These planes are easily determined using Mohr’s Circle.

REMARK
The two planes form a 45º
angle with the principal
stress directions and, thus,
are orthogonal to each
other. They form an angle
of 45º with the isostatics.

42
Equation of the maximum shear lines
 Consider the general equation of a slip line y = f ( x ), the relation
2τ xy π  π 1
tan 2α = and β= α + tan ( 2 β ) =
tan  2α −  = −
σ x −σ y 4  2 tan 2α
 Then,
1 σ x −σ y 2 tan β
tan ( 2 β ) =
− =
− =2
tan ( 2α ) 2τ xy 1 − tan β σ x −σ y 2 y′
− =2
tan (=
dy not
β ) = y′
2τ xy 1 − ( y′ )
dx

4τ xy
( y′ ) − y′ − 1 =0
2

σ x −σ y

43
Equation of the maximum shear lines
 Solving the 2nd order eq.: φ ( x, y ) Known this function,
the eq. can be integrated
Differential 2 to obtain a family of
2τ xy  2τ xy 
equation of the y'= ± 
 σ −σ   + 1 curves of the type:
slip lines σ x −σ y  x y 
=y f ( x) + C

44
Chapter 7
Plane Linear Elasticity

rs
ee
s gin
7.1 Introduction

t d le En

r
As seen in Chapter 6, from a mathematical point of view, the elastic problem

ba
ge ro or
eS m
consists in a system of PDEs that must be solved  in the three dimensions of

ci
space and in the dimension associated with time R3 × R+ . However, in certain
f

ra
C d P cs
situations, the problem can be simplified so that it is reduced
 to two dimensions
b
a
in space in addition to, obviously, the temporal dimension R2 × R+ . This sim-
i
an an n

plification is possible because, in certain cases, the geometry and boundary con-
y ha

ditions of the problem allow identifying an irrelevant direction (associated with


le
liv or ec

a direction of the problem) such that solutions independent of this dimension can
be posed a priori for this elastic problem.
M

.A

Consider a local coordinate system {x, y, z} in which the aforementioned ir-


m

relevant direction (assumed constant) coincides with the z-direction. Then, the
d

analysis is reduced to the x-y plane and, hence, the name plane elasticity used to
uu
e

denote such problems. In turn, these are typically divided into two large groups
X Th

er
tin

associated with two families of simplifying hypotheses, plane stress problems


and plane strain problems.
on

.O

For the sake of simplicity, the isothermal case will be considered here, even
C

though there is no intrinsic limitation to generalizing the results that will be


©

obtained to the thermoelastic case.

7.2 Plane Stress State


The plane stress state is characterized by the following simplifying hypotheses:
1) The stress state is of the type
⎡ ⎤
σx τxy 0
not ⎢ ⎥
σ ]xyz ≡ ⎢
[σ ⎣ τxy σy 0 ⎥
⎦. (7.1)
0 0 0

339
340 C HAPTER 7. P LANE L INEAR E LASTICITY

2) The non-zero stresses (that is, those associated with the x-y plane) do not
depend on the z-variable,

σx = σx (x, y,t) , σy = σy (x, y,t) and τxy = τxy (x, y,t) . (7.2)

To analyze under which conditions these hypotheses are reasonable, consider


a plane elastic medium whose dimensions and form associated with the x-y plane
(denoted as plane of analysis) are arbitrary and such that the third dimension (de-
noted as the thickness of the piece) is associated with the z-axis (see Figure 7.1).
Assume the following circumstances hold for this elastic medium:

rs
ee
a) The thickness e is much smaller than the typical dimension associated with

s gin
the plane of analysis x-y,
eL. (7.3)

t d le En
b) The actions (body forces b (x,t), prescribed displacements u∗ (x,t) and trac-

r
ba
ge ro or
tion vector t∗ (x,t) ) are contained within the plane of analysis x-y (its z-

eS m
ci
f
component is null) and, in addition, do not depend on the third dimension,

ra
C d P cs
⎡ ⎤ ⎡ ⎤
b
a
bx (x, y,t) u∗x (x, y,t)
i
an an n

not ⎢ ⎥ ⎢ ⎥
b≡⎢ ⎥, ⎢ u∗ (x, y,t) ⎥ ,
y ha

Γ ∗ not

⎣ by (x, y,t) ⎦ u : u ⎣ y ⎦
le
liv or ec

0 −
M

.A

⎡ ⎤ (7.4)
tx∗ (x, y,t)
m



not ⎢ ∗ ⎥
d

Γσ = Γσ+ Γσ− Γσe : t∗ ≡ ⎢ t (x, y,t) ⎥.


uu

⎣y ⎦
e
X Th


er
tin

c) The traction vector t∗ (x,t) is only non-zero on the boundary of the piece’s
on

.O

thickness (boundary Γσe ), whilst on the lateral surfaces Γσ+ and Γσ− it is null
C

(see Figure 7.1).


©

⎡ ⎤
0
+
⎢ ⎥
− ∗ not ⎢ ⎥
Γσ Γσ : t ≡ ⎣ 0 ⎦ . (7.5)
0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Plane Stress State 341

rs
ee
s gin
Figure 7.1: Example of a plane stress state.

t d le En

r
ba
ge ro or
eS m
ci
f

ra
Remark 7.1. The piece with the actions defined by (7.4) and (7.5) is
C d P cs
b
a
compatible with the plane stress state given by (7.1) and (7.2), and
i
an an n

schematized in Figure 7.21 . In effect, applying the boundary condi-


y ha

tions Γσ on the piece yields:


le
liv or ec

• Lateral surfaces Γσ+ and Γσ−


M

.A

⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ ⎤
0 σx τxy 0 0 0
m

not ⎢ ⎥ not ⎢ ⎥⎢ ⎥ ⎢ ⎥
d

n≡⎣ 0 ⎦ , σ · n ≡ ⎣ τxy σy 0 ⎦ ⎣ 0 ⎦ = ⎣ 0 ⎦ ,
uu
e

±1 ±1
X Th

0 0 0 0
er
tin

• Edge Γσe
on

.O

⎡ ⎤ ⎡ ⎤⎡ ⎤ ⎡ ⎤
C

nx σx τxy 0 nx tx (x, y,t)


©

not ⎢ ⎥ not ⎢ ⎥⎢ ⎥ ⎢
n ≡ ⎣ ny ⎦ , σ (x, y,t)·n ≡ ⎢ τ σ 0 ⎥ ⎣ ny ⎦ = ⎣ ty (x, y,t) ⎥
⎦,
⎣ xy y ⎦
0 0 0 0 0 0

which is compatible with the assumptions (7.4) and (7.5) .

1 The fact that all the non-null stresses are contained in the x-y plane is what gives rise to the
name plane stress.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
342 C HAPTER 7. P LANE L INEAR E LASTICITY

rs
ee
Figure 7.2: Plane stress state.

s gin
t d le En

r
ba
ge ro or
7.2.1 Strain Field. Constitutive Equation

eS m
ci
f

ra
Consider now the linear elastic constitutive equation (6.24),
C d P cs
b
a
i
ν 1+ν ν 1
an an n

ε =− σ)1+
Tr (σ σ)1+
σ = − Tr (σ σ, (7.6)
y ha

E E E 2G
le
liv or ec

which, applied on the stress state in (7.1) and in engineering notation, provides
the strains (6.25)2
M

.A
m

1 1 1
d

εx = (σx − ν (σy + σz )) = (σx − νσy ) γxy = τxy ,


uu
e

E E G
X Th

er
tin

1 1 1 (7.7)
εy = (σy − ν (σx + σz )) = (σy − νσx ) γxz = τxz = 0 ,
E E G
on

.O

1 ν 1
C

εz = (σz − ν (σx + σy )) = − (σx + σy ) γyz = τyz = 0 ,


E E G
©

where the conditions σz = τxz = τyz = 0 have been taken into account. From (7.2)
and (7.7) it is concluded that the strains do not depend on the z-coordinate either
(εε = ε (x, y,t)). In addition, the strain εz in (7.7) can be solved as
ν
εz = − (εx + εy ) . (7.8)
1−ν

2 The engineering angular strains are defined as γxy = 2 εxy , γxz = 2 εxz and γyz = 2 εyz .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Plane Stress State 343

In short, the strain tensor for the plane stress case results in

⎡ 1 ⎤
εx γxy 0
⎢ 2 ⎥ ν
not ⎢ ⎥
ε (x, y,t) ≡ ⎢ 1 γxy εy 0 ⎥ with εz = − (εx + εy ) (7.9)
⎣2 ⎦ 1−ν
0 0 εz

and replacing (7.8) in (7.7) leads, after certain algebraic operations, to

rs
ee
E E
σx = (εx + νεy ) , σy = (εy + νεx ) ,

s gin
1 − ν2 1 − ν2 (7.10)
E

t d le En
and τxy = γxy ,
2 (1 + ν)

r
ba
ge ro or
eS m
which can be rewritten as

ci
⎡ ⎤⎡ ⎤
⎡ ⎤ f

ra
C d P cs
σx 1 ν
⎥ ⎢ εx ⎥
0
b
a
⎢ ⎥ ⎢
i
⎢σ ⎥= E ⎢ ⎥⎢ ⎥
an an n

plane
⎣ y ⎦ 1 − ν2 ⎢ ν 1 0 ⎥ ⎣ εy ⎦ =⇒ σ } = C stress · {εε } .

⎣ ⎦
y ha

1−ν
τxy γxy
le
0 0
liv or ec

   2   
M

σ}
{σ {εε }
.A

plane
C stress
(7.11)
m

d
uu
e
X Th

7.2.2 Displacement Field


er
tin

The components of the geometric equation of the problem (6.3),


on

.O

1
C

ε (x,t) = ∇S u (x,t) = (u ⊗ ∇ + ∇ ⊗ u) , (7.12)


©

2
can be decomposed into two groups:
1) Those that do not affect the displacement uz (and are hypothetically inte-
grable in R2 for the x-y domain),

∂ ux ⎪

εx (x, y,t) = ⎪
⎪ integration
∂x ⎪
⎪ 
⎬ in R2 ux = ux (x, y,t)
∂ uy =⇒ . (7.13)
εy (x, y,t) = ⎪ uy = uy (x, y,t)
∂y ⎪

∂ ux ∂ uy ⎪



γxy (x, y,t) = 2εxy = +
∂y ∂x

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
344 C HAPTER 7. P LANE L INEAR E LASTICITY

2) Those in which the displacement uz intervenes,


∂ uz ν
εz (x, y,t) = =− (εx + εy ) ,
∂z 1−ν
∂ ux ∂ uz (7.14)
γxz (x, y,t) = 2εxz = + =0,
∂z ∂x
∂ uy ∂ uz
γyz (x, y,t) = 2εyz = + =0.
∂z ∂y

rs
Observation of (7.1) to (7.14) suggests considering an ideal elastic plane

ee
stress problem reduced to the two dimensions of the plane of analysis and char-

n
acterized by the unknowns

gi
t d le En
   
  εx σx
u
u (x, y,t) ≡ x , {εε (x, y,t)} ≡  εy  and {σ
not not not
σ (x, y,t)} ≡  σy  , (7.15)

ar
ar s
uy

ge ro or
eS m

ib
γxy τxy

ac
f
C d P cs
b
in which the additional unknowns with respect to the general problem are either
i
null, or can be calculated in terms of those in (7.15), or do not intervene in the
an an n
y ha

reduced problem,
ν
le
σz = τxz = τyz = γxz = γyz = 0 , εz = − (εx + εy ) ,
liv or ec

1−ν (7.16)
M

.A

and uz (x, y, z,t) does not intervene in the problem.


m

d
uu
e

Remark 7.2. The plane stress problem is an ideal elastic problem


X Th

er
tin

since it cannot be exactly reproduced as a particular case of a three-


dimensional elastic problem. In effect, there is no guarantee that the
on

.O

solution of the reduced plane stress ux (x, y,t) and uy (x, y,t) will al-
low obtaining a solution uz (x, y, z,t) for the rest of components of
C

the geometric equation (7.14).

7.3 Plane Strain


The strain state is characterized by the simplifying hypotheses
   
ux ux (x, y,t)
not
u ≡  uy  =  uy (x, y,t)  . (7.17)
uz 0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Plane Strain 345

Again, it is illustrative to analyze in which situations these hypotheses are plau-


sible. Consider, for example, an elastic medium whose geometry and actions
can be generated from a bidimensional section (associated with the x-y plane
and with the actions b (x,t), u∗ (x,t) and t∗ (x,t) contained in this plane) that
is translated along a straight generatrix perpendicular to said section and, thus,
associated with the z-axis (see Figure 7.3).
The actions of the problem can then be characterized by
⎡ ⎤ ⎡ ∗ ⎤ ⎡ ∗ ⎤
bx (x, y,t) ux (x, y,t) tx (x, y,t)
not ⎢ ⎥ not ⎢ ⎥ not ⎢ ⎥
b ≡ ⎣ by (x, y,t) ⎦ , Γu : u∗ ≡ ⎣ u∗y (x, y,t) ⎦ and Γσ : t∗ ≡ ⎣ ty∗ (x, y,t) ⎦ . (7.18)

rs
0 0 0

ee
s gin
In the central section (which is a plane of symmetry with respect to the z-axis)
the conditions

t d le En
∂ ux ∂ uy
uz = 0 , =0 =0

r
and (7.19)

ba
∂z ∂z

ge ro or
eS m
ci
f

ra
are satisfied and, thus, the displacement field in this central section is of the form
C d P cs
b
a
⎡ ⎤
i
ux (x, y,t)
an an n

not ⎢ ⎥
y ha

u (x, y,t) ≡ ⎣ uy (x, y,t) ⎦ . (7.20)


le
liv or ec

0
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 7.3: Example of a plane strain state.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
346 C HAPTER 7. P LANE L INEAR E LASTICITY

7.3.1 Strain and Stress Fields


The strain field corresponding with the displacement field characteristic of a
plane strain state (7.20) is

∂ ux ∂ uz
εx (x, y,t) = , εz (x, y,t) = =0,
∂x ∂z
∂ uy ∂ ux ∂ uz
εy (x, y,t) = , γxz (x, y,t) = + =0, (7.21)
∂y ∂z ∂x
∂ ux ∂ uy ∂ u y ∂ uz

rs
γxy (x, y,t) = + , γyz (x, y,t) = + =0.
∂y ∂x ∂z ∂y

ee
s gin
Therefore, the structure of the strain tensor is3

t d le En
⎡ 1 ⎤
εx γxy 0

r
⎢ 2 ⎥

ba
ge ro or
not ⎢ ⎥

eS m
ε (x, y,t) ≡ ⎢ 1 γxy εy 0 ⎥ . (7.22)

ci
⎣2 ⎦
f

ra
C d P cs
b
a
0 0 0
i
an an n
y ha

le
Consider now the lineal elastic constitutive equation (6.20)
liv or ec
M

.A

σ = λ Tr (εε ) 1 + 2μεε = λ Tr (εε ) 1 + 2Gεε , (7.23)


m

which, applied to the strain field (7.21), produces the stresses


d
uu
e
X Th

er

σx = λ (εx + εy ) + 2μεx = (λ + 2G) εx + λ εy , τxy = G γxy ,


tin

σy = λ (εx + εy ) + 2μεy = (λ + 2G) εy + λ εx , τxz = G γxz = 0 ,


on

(7.24)
.O
C

σz = λ (εx + εy ) , τyz = G γyz = 0 .


©

Considering (7.21) and (7.24), one concludes that stresses do not depend on the
σ = σ (x, y,t)). On the other hand, the stress σz in (7.24)
z-coordinate either (σ
can be solved as
λ
σz = (σx + σy ) = ν (σx + σy ) (7.25)
2 (λ + μ)

3 By analogy with the plane stress case, the fact that all non-null strains are contained in the
x-y plane gives rise to the name plane strain.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Plane Strain 347

and the stress tensor for the plane strain case results in

⎡ ⎤
σx τxy 0
not ⎢ ⎥
σ (x, y,t) ≡ ⎢ ⎣ τxy σy 0 ⎥⎦ with σz = −ν (σx + σy ) , (7.26)
0 0 σz

where the non-null components of the stress tensor (7.26) are

rs
 

ee
E (1 − ν) ν
σx = (λ + 2G) εx + λ εy = εx + εy ,

s gin
(1 + ν) (1 − 2ν) 1−ν
 

t d le En
E (1 − ν) ν (7.27)
σy = (λ + 2G) εy + λ εx = εy + εx ,
(1 + ν) (1 − 2ν) 1−ν

r
ba
ge ro or
eS m
ci
E
and τxy = G γxy = f
γxy .

ra
2 (1 + ν)
C d P cs
b
a
i
an an n

Equation (7.27) can be rewritten in matrix form as


y ha

⎡ ν ⎤
⎡ ⎤ ⎡ ⎤
le
liv or ec

1 0
σx ⎢ 1−ν ⎥ εx
⎢ ⎥ (1 − ν) ⎢ ν ⎥ ⎢ ⎥
M

.A

⎢ σy ⎥ = E ⎢ ⎥ ⎢ εy ⎥ ⇒
1 0
⎣ ⎦ (1 + ν) (1 − 2ν) ⎢ 1 − ν ⎥ ⎣ ⎦
⎣ 1 − 2ν ⎦
m

τxy γxy
d

0 0
2 (1 − ν)
uu

   
e

  (7.28)
X Th

σ}
{σ {εε }
er
tin

plane
C strain
on

.O
C

plane
σ } = C strain · {εε } .

©

Similarly to the plane stress problem, (7.20), (7.21) and (7.26) suggest con-
sidering an elastic plane strain problem reduced to the two dimensions of the
plane of analysis x-y and characterized by the unknowns
⎡ ⎤ ⎡ ⎤
  εx σx
not ux not ⎢ ⎥ not ⎢ ⎥
u (x, y,t) ≡ σ (x, y,t)} ≡ ⎣ σy ⎦ ,
, {εε (x, y,t)} ≡ ⎣ εy ⎦ and {σ (7.29)
uy
γxy τxy

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
348 C HAPTER 7. P LANE L INEAR E LASTICITY

rs
ee
s gin
Figure 7.4: The plane linear elastic problem.

t d le En

r
ba
ge ro or
eS m
in which the additional unknowns with respect to the general problem are either

ci
f
null or can be calculated in terms of those in (7.29),

ra
C d P cs
b
a
uz = 0 , εz = γxz = γyz = τxz = τyz = 0 and σz = ν (σx + σy ) . (7.30)
i
an an n
y ha

le
liv or ec

7.4 The Plane Linear Elastic Problem


M

.A

In view of the equations in Sections 7.2 and 7.3, the linear elastic problem for
m

the plane stress and plane strain problems is characterized as follows (see Fig-
uu
e

ure 7.4).
X Th

er
tin

Equations 4
on

.O

a) Cauchy’s equation
C

∂ σx ∂ τxy ∂ 2 ux
+ + ρbx = ρ 2
∂x ∂y ∂t (7.31)
∂ τxy ∂ σy ∂ 2 uy
+ + ρby = ρ 2
∂x ∂y ∂t

4 The equation corresponding to the z-component either does not intervene (plane stress), or
is identically null (plane strain).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
The Plane Linear Elastic Problem 349

b) Constitutive equation
⎡ ⎤ ⎡ ⎤
σx εx
not ⎢ ⎥ not ⎢ ⎥
σ } ≡ ⎣ σy ⎦ ,
{σ {εε } ≡ ⎣ εy ⎦ ; σ } = C · {εε } ,
{σ (7.32)
τxy γxy

where the constitutive matrix C can be written in a general form, from (7.11)
and (7.28), as

Ē = E

rs
⎡ ⎤ Plane stress

ee
1 ν̄ 0 ν̄ = ν
⎢ ⎥ ⎧

s gin
not Ē ⎢ ⎥
C≡ ⎢ ν̄ 1 0 ⎥ ⎪
⎪ E (7.33)
1 − ν̄ 2 ⎣ ⎨ Ē =
1 − ν̄ ⎦

t d le En
0 0 Plane strain 1−ν 2

⎩ ν̄ = ν

2

r
ba
ge ro or
1−ν

eS m
ci
f

ra
c) Geometric equation
C d P cs
b
a
i
an an n

∂ ux ∂ uy ∂ ux ∂ uy
y ha

εx = , εy = , γxy = + (7.34)
∂x ∂y ∂y ∂x
le
liv or ec
M

.A

d) Boundary conditions in space


m

d
uu

   
e

u∗x (x, y,t) tx∗ (x, y,t)


X Th

er

Γu : u∗ ≡ Γσ : t∗ ≡
tin

not not
,
u∗y (x, y,t) ty∗ (x, y,t)
on

.O

(7.35)
   
C

σx τxy nx
©

t∗ = σ · n ,
not not
σ≡ , n≡
τxy σy ny

e) Initial conditions
 
 . 
u (x, y,t)  =0, u (x, y,t)  = v0 (x, y) (7.36)
t=0 t=0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
350 C HAPTER 7. P LANE L INEAR E LASTICITY

Unknowns

⎡ ⎤
  1  
not ux not ⎢
εx γxy ⎥ σ τ
x xy
u (x, y,t) ≡ , ε (x, y,t) ≡ ⎣ 1 2 ⎦, σ (x, y,t) not
≡ (7.37)
uy γxy εy τ xy σy
2

Equations (7.31) to (7.37) define a system of 8 PDEs with 8 unknowns that


must be solved in the reduced space-time domain R2 × R+ . Once the problem
is solved, the following can be explicitly calculated:

rs
ee
ν
Plane stress → εz = (εx + εy )
1−ν

s gin
(7.38)
Plane strain → σz = ν (σx + σy )

t d le En

r
ba
ge ro or
eS m
7.5 Problems Typically Assimilated to Plane Elasticity

ci
f

ra
C d P cs
7.5.1 Plane Stress
b
a
i
an an n

The stress and strain states produced in solids that have a dimension consider-
y ha

ably inferior to the other two (which constitute the plane of analysis x-y) and
le
liv or ec

whose actions are contained in said plane are typically assimilated to a plane
stress state. The slab loaded on its mean plane and the deep beam of Figure 7.5
M

.A

are classic examples of structures that can be analyzed as being in a plane stress
m

state. As a particular case, the problems of simple and complex bending in beams
d
uu

considered in strength of materials can also be assimilated to plane stress prob-


e

lems.
X Th

er
tin
on

.O
C

Figure 7.5: Slab loaded on its mean plane (left) and deep beam (right).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems Typically Assimilated to Plane Elasticity 351

7.5.2 Plane Strain


The solids whose geometry can be obtained by translation of a plane section
with actions contained in its plane (plane of analysis x-y) along a generatrix line
perpendicular to said section are typically assimilated to plane strain states. In
addition, the plane strain hypothesis εz = γxz = γyz = 0 must be justifiable. In
general, this situation occurs in two circumstances:
1) The dimension of the piece in the z-direction is very large (for the purposes
of analysis, it is assumed to be infinite). In this case, any central transversal
section (not close to the extremes) can be considered a symmetry plane and,

rs
thus, satisfies the conditions

ee
∂ ux ∂ uy
uz = 0 , = 0 and =0, (7.39)

s gin
∂z ∂z
which result in the initial condition of the plane strain state (7.17),

t d le En
⎡ ⎤ ⎡ ⎤
ux ux (x, y,t)

r
ba
not ⎢ ⎥ ⎢ ⎥

ge ro or
eS m
u ≡ ⎣ uy ⎦ = ⎣ uy (x, y,t) ⎦ . (7.40)

ci
f

ra
uz 0
C d P cs
b
a
i
Examples of this case are a pipe under internal (and/or external) pressure
an an n
y ha

(see Figure 7.6), a tunnel (see Figure 7.7) and a strip foundation (see Fig-
ure 7.8).
le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 7.6: Pressure tube.

2) The length of the piece in the longitudinal direction is reduced, but the dis-
placements in the z-direction are impeded by the boundary conditions at the
end sections (see Figure 7.9).
In this case, the plane strain hypothesis (7.17) can be assumed for all the
transversal sections of the piece.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
352 C HAPTER 7. P LANE L INEAR E LASTICITY

rs
ee
Figure 7.7: Tunnel.

s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e

Figure 7.8: Strip foundation.


X Th

er
tin
on

.O
C

Figure 7.9: Solid with impeded z-displacements.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Representative Curves of Plane Elasticity 353

7.6 Representative Curves of Plane Elasticity


There is an important tradition in engineering of graphically representing the
distribution of plane elasticity. To this aim, certain families of curves are used,
whose plotting on the plane of analysis provides useful information of said stress
state.

7.6.1 Isostatics or stress trajectories

rs
Definition 7.1. The isostatics or stress trajectories are the envelopes

ee
of the vector field determined by the principal stresses.

s gin
t d le En
Considering the definition of the envelope of a vector field, isostatics are, at each

r
point, tangent to the two principal directions and, thus, there exist two families

ba
ge ro or
eS m
of isostatics:

ci
f
− Isostatics σ1 , tangent to the direction of the largest principal stress.

ra
C d P cs
− Isostatics σ2 , tangent to the direction of the smallest principal stress.
b
a
i
an an n

In addition, since the principal stress directions are orthogonal to each other,
y ha

both families of curves are also be orthogonal. The isostatic lines provide infor-
le
mation on the mode in which the flux of principal stresses occurs on the plane
liv or ec

of analysis.
M

.A

As an example, Figure 7.10 shows the distribution of isostatics on a supported


beam with uniformly distributed loading.
m

d
uu
e
X Th

er

Definition 7.2. A singular point is a point characterized by the stress


tin

state
σx = σy and τxy = 0
on

.O

and its Mohr’s circle is a point on the axis σ (see Figure 7.11).
C

A neutral point is a singular point characterized by the stress state


σx = σy = τxy = 0

and its Mohr’s circle is the origin of the σ − τ space (see Fig-
ure 7.11).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
354 C HAPTER 7. P LANE L INEAR E LASTICITY

Figure 7.10: Isostatics or stress trajectories on a beam.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
Figure 7.11: Singular and neutral points.

ci
f

ra
C d P cs
b
a
i
an an n
y ha

Remark 7.3. All directions in a singular point are principal stress di-
le
rections (the pole is the Mohr’s circle itself, see Figure 7.11). Conse-
liv or ec

quently, the isostatics tend to loose their regularity in singular points


M

.A

and can brusquely change their direction.


m

d
uu
e
X Th

er
tin

7.6.1.1 Differential Equation of the Isostatics


on

.O

Consider the general equation of an isostatic line y = f (x) and the value of the
angle formed by the principal stress direction σ1 with respect to the horizontal
C

direction (see Figure 7.12),


©


2τxy 2 tan α ⎪⎪

tan (2α) = = ⎬ 
σx − σy 1 − tan2 α ⇒ 2τxy = 2y ⇒
dy not  ⎪
⎪ σx − σy 1 − (y )2
tan α = =y ⎪
⎭ (7.41)
dx

σx − σy 
(y )2 + y −1 = 0
τxy

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Representative Curves of Plane Elasticity 355

Figure 7.12: Determination of the differential equation of the isostatics.

rs
ee
and solving the second-order equation (7.41) for y , the differential equation of

s gin
the isostatics is obtained.

t d le En
 
Differential σx − σy 2

r
σ − σ

ba
 x y

ge ro or
→ y =− ± +1

eS m
equation (7.42)
2τxy 2τxy

ci
of the isostatics  f 

ra
C d P cs
ϕ (x, y)
b
a
i
an an n
y ha

If the function ϕ (x, y) in (7.42) is known, this equation can be integrated to


le
obtain the algebraic equation of the family of isostatics,
liv or ec
M

.A

y = f (x) +C . (7.43)
m

The double sign in (7.42) leads to two differential equations corresponding to


d
uu

the two families of isostatics.


e
X Th

er
tin

Example 7.1 – A rectangular plate is subjected to the following stress states.


on

.O

σx = −x3 ; σy = 2x3 − 3xy2 ; τxy = 3x2 y ; τxz = τyz = σz = 0


C

Obtain and plot the singular points and distribution of isostatics.

Solution
The singular points are defined by σx = σy and τxy = 0 . Then,
⎧ 

⎪ σx = −x3 = 0



⎨ x = 0 =⇒ ∀y
σy = 2x3 − 3xy2 = 0
τxy = 3x y = 0 =⇒
2




⎪ σx = −x3

⎩ y = 0 =⇒ =⇒ x = 0
σy = 2x3 − 3xy2 = 2x3

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
356 C HAPTER 7. P LANE L INEAR E LASTICITY

Therefore, the locus of singular points is the straight line x = 0. These singu-
lar points are, in addition, neutral points (σx = σy = 0).
The isostatics are obtained from (7.42),
 2
 dy σx − σy σx − σy
y = =− ± +1 ,
dx 2τxy 2τxy

which, for the given data of this problem, results in



rs

⎪ dy x
⎨ = x2 − y2 = C1

ee
dx y =⇒ integrating =⇒ .
⎪ xy = C2

s gin

⎪ dy −y
⎩ =
dx x

t d le En
Therefore, the isostatics are two families of equilateral hyperboles orthogo-

r
ba
ge ro or
nal to each other.

eS m
ci
f
On the line of singular points x = 0 (which divides the plate in two regions)

ra
C d P cs
the isostatics will brusquely change their slope. To identify the family of
b
a
isostatics σ1 , consider a point in each region:
i
an an n
y ha

− Point (1, 0): σx = σ2 = −1; σy = σ1 = +2; τxy = 0


le
(isostatic σ1 in the y-direction)
liv or ec

− Point (−1, 0): σx = σ1 = +1; σy = σ2 = −2; τxy = 0


M

.A

(isostatic σ1 in the x-direction)


m

Finally, the distribution of isostatics is as follows:


uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Representative Curves of Plane Elasticity 357

7.6.2 Isoclines

Definition 7.3. Isoclines are the locus of the points in the plane of
analysis along which the principal stress directions form a certain
angle with the x-axis.

It follows from its definition that in all the points of a same isocline the principal

rs
stress directions are parallel to each other, forming a constant angle θ (which

ee
characterizes the isocline) with the x-axis (see Figure 7.13).

s gin
7.6.2.1 Equation of the Isoclines

t d le En
The equation y = f (x) of the isocline with an angle θ is obtained by establishing
that the principal stress direction σ1 forms an angle α = θ with the horizontal

r
ba
ge ro or
direction, that is,

eS m
ci
f

ra
C d P cs
b
a
Algebraic equation 2τxy
i
tan (2θ ) =
an an n

of the isoclines σx − σy (7.44)


y ha

 
le
ϕ (x, y)
liv or ec
M

.A

This algebraic equation allows isolating, for each value of θ ,


m

y = f (x, θ ) , (7.45)
uu
e
X Th

er
tin

which constitutes the equation of the family of isoclines parametrized in terms


of the angle θ .
on

.O
C

Figure 7.13: Isocline.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
358 C HAPTER 7. P LANE L INEAR E LASTICITY

Remark 7.4. Determining the family of isoclines allows knowing, at


each point in the medium, the direction of the principal stresses and,
thus, the obtainment of the isostatics can be sought. Given that iso-
clines can be determined by means of experimental methods (meth-
ods based on photoelasticity) they provide, indirectly, a method for
the experimental determination of the isostatics.

rs
ee
7.6.3 Isobars

s gin
t d le En
Definition 7.4. Isobars are the locus of points in the plane of analy-

r
ba
sis with the same value of principal stress σ1 (or σ2 ).

ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
Two families of isobars will cross at each point of the plane of analysis: one
an an n

corresponding to σ1 and another to σ2 . Note that the isobars depend on the value
y ha

of σ1 , but not on its direction (see Figure 7.14).


le
liv or ec
M

.A

7.6.3.1 Equation of the Isobars


m

The equation that provides the value of the principal stresses (see Chapter 4) im-
d

plicitly defines the algebraic equation of the two families of isobars y = f1 (x, c1 )
uu
e

and y = f2 (x, c2 ),
X Th

er
tin

⎧ 

⎪ 2
on

.O


⎪ σ + σ σ − σ

⎪ σ1 =
x y
+
x y
+ τxy
2 = const. = c

⎪ 1
C

⎪ 2 2
⎪  
Algebraic ⎪
©



equation ϕ1 (x, y)
 (7.46)
of the ⎪⎪  
isobars ⎪⎪
⎪ σx + σ y σx − σy 2

⎪ σ2 = − + τxy
2 = const. = c

⎪ 2 2
2

⎪  


ϕ2 (x, y)

which leads to 
y = f1 (x, c1 )
(7.47)
y = f2 (x, c2 )

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Representative Curves of Plane Elasticity 359

Figure 7.14: Isobars.

rs
ee
s gin
7.6.4 Maximum Shear Stress or Slip Lines

t d le En

r
ba
ge ro or
eS m
Definition 7.5. Maximum shear stress lines or slip lines are the en-

ci
f
velopes of the directions that, at each point, correspond with the

ra
C d P cs
b
maximum value (in modulus) of the shear (or tangent) stress.

a
i
an an n
y ha

le
liv or ec
M

.A

Remark 7.5. At each point of the plane of analysis there are two
planes on which the shear stresses reach the same maximum value
m

(in module) but that have opposite directions, τmax and τmin . These
d
uu

planes can be determined by means of the Mohr’s circle and form


e
X Th

a 45◦ angle with the principal stress directions (see Figure 7.15).
er
tin

Therefore, their envelopes (maximum shear stress lines) are two


families of curves orthogonal to each other that form a 45◦ angle
on

.O

with the isostatics.


C

7.6.4.1 Differential Equation of the Maximum Shear Lines


Consider β is the angle formed by the direction of τmax with the horizontal
direction (see Figure 7.16). In accordance with Remark 7.55 ,
π π! 1
β =α− =⇒ tan (2β ) = tan 2α − =− , (7.48)
4 2 tan (2α)

5 Here, the trigonometric expression tan (θ − π/2) = − cot θ = −1/tan θ is used.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
360 C HAPTER 7. P LANE L INEAR E LASTICITY

rs
ee
Figure 7.15: Maximum shear stress planes.

s gin
t d le En
where α is the angle formed by the principal stress direction σ1 with the

r
ba
horizontal direction. Consequently, considering the general equation of a slip

ge ro or
eS m
line, y = f (x), the expression (7.48) and the relation tan (2α) = 2τxy / (σx − σy )

ci
f

ra
yields
C d P cs

b
a
1 σx − σy 2 tan β ⎪
i
tan (2β ) = − = = ⎬
an an n

tan (2α) 2τxy 1 − tan2 β =⇒


y ha

dy not  ⎪

tan β = =y
le
liv or ec

dx (7.49)
M

.A

σx − σy 2y 4τxy
=⇒ (y )2 − y − 1 = 0.
m

− =
d

2τxy 1 − (y )2 σx − σy
uu
e
X Th

Solving the second-order equation in (7.49) for y provides the differential equa-
er
tin

tion of the maximum shear stress lines.


on

.O


C

Differential 2
2τ 2τxy
©

equation of the  xy
y =− ± +1 (7.50)
max. shear stress σx − σy σx − σy
or slip lines  
ϕ (x, y)

If the function ϕ (x, y) in (7.50) is known, this differential equation can be in-
tegrated and the algebraic equation of the two families of orthogonal curves
(corresponding to the double sign in (7.50)) is obtained.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Representative Curves of Plane Elasticity 361

Figure 7.16: Maximum shear stress or slip lines.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
362 C HAPTER 7. P LANE L INEAR E LASTICITY

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 363

P ROBLEMS

Problem 7.1 – Justify whether the following statements are true or false.
a) If a plane stress state has a singular point, all the isoclines cross this
point.
b) If a plane stress state is uniform, all the slip lines are parallel to each

rs
other.

ee
s gin
Solution

t d le En
a) A singular point is defined as:

r
ba
ge ro or
eS m
ci
f

ra
"
C d P cs
The stress state is
b
a
σ1 = σ2 represented by a point.
i
an an n

τ =0
y ha

le
liv or ec
M

.A

Therefore, all directions are principal stress directions and, given an angle θ
which can take any value, the principal stress direction will form an angle θ
m

with the x-axis. Then, an isocline of angle θ will cross said point and, since this
uu
e

holds true for any value of θ , all the isoclines will cross this point. Therefore,
X Th

er
tin

the statement is true.


on

.O

b) A uniform stress state implies that the Mohr’s circle is equal in all points of
the medium, therefore, the planes of maximum shear stress will be the same in
C

all points. Then, the maximum shear stress lines (or slip lines) will be parallel to
©

each other. In conclusion, the statement is true.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
364 C HAPTER 7. P LANE L INEAR E LASTICITY

Problem 7.2 – A rectangular plate is subjected to the following plane stress


states.

1) σx = 0 ; σy = b > 0 ; τxy = 0
2) σx = 0 ; σy = 0 ; τxy = m y , m > 0

Plot for each state the isostatics and the slip lines, and indicate the singular
points.

rs
ee
s gin
Solution

t d le En
1) The Mohr’s circle for the stress state σx = 0 ; σy = b > 0 ; τxy = 0 is:

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Then, the isostatics are:


©

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 365

And the slip lines are:

rs
There do not exist singular points for this stress state.

ee
s gin
t d le En
2) The Mohr’s circle for the stress state σx = 0 ; σy = 0 ; τxy = m y , m > 0 is:

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
366 C HAPTER 7. P LANE L INEAR E LASTICITY

Then, the isostatics and singular points are:

And the slip lines are:

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 367

E XERCISES

7.1 – A rectangular plate is subjected to the following plane strain state:

σx = σy
τxy = ax
σy = b

rs
(a > 0 , b > 0)

ee
Plot the isostatics and the slip lines, and indicate the singular points.

s gin
t d le En
7.2 – Plot the isostatics in the transversal section of the cylindrical shell shown

r
below. Assume a field of the form:

ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i

an an n

⎪ B
y ha

⎨ ur = Ar + r ;
⎪ A > 0, B > 0
le
liv or ec

⎪ uθ = 0


M

.A

uz = 0
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
CH.8. PLASTICITY
Multimedia Course on Continuum Mechanics
Overview
 Introduction Lecture 1
 Previous Notions Lecture 2 Lecture 3
 Principal Stress Space
Lecture 4
 Normal and Shear Octahedral Stresses
 Stress Invariants
Lecture 5
 Effective Stress
 Rheological Friction Models
 Elastic Element
 Frictional Element
 Elastic-Frictional Model

2
Overview (cont’d)
 Rheological Friction Models (cont’d)
 Frictional Model with Hardening
 Elastic-Frictional Model with Hardening
 Phenomenological Behaviour
 Notion of Plastic Strain
 Notion of Hardening Lecture 6
 Bauschinger Effect
 Elastoplastic Behaviour
 1D Incremental Theory of Plasticity
 Additive Decomposition of Strain Lecture 7
 Hardening Variable
 Yield Stress, Yield Function and Space of Admissible Stresses Lecture 8 Lecture 9
 Constitutive Equation Lecture 9
 Elastoplastic Tangent Modulus
Lecture 10
 Uniaxial Stress-Strain Curve
3
Overview (cont’d)
 3D Incremental Theory of Plasticity
 Additive Decomposition of Strain
 Hardening Variable Lecture 11
 Yield Function
 Loading - Unloading Conditions and Consistency Conditions
 Constitutive Equation
Lecture 12
 Elastoplastic Constitutive Tensor
 Yield Surfaces
Lecture 13
 Von Mises Criterion
 Tresca Criterion
 Mohr-Coulomb Criterion Lecture 14
 Drucker-Prager Criterion

4
8.1 Introduction
Ch.8. Plasticity

5
Introduction
 A material with plastic behavior is characterized by:
 A nonlinear stress-strain relationship.
 The existence of permanent (or plastic) strain during a
loading/unloading cycle.
 Lack of unicity in the stress-strain relationship.

 Plasticity is seen in most materials, after an initial elastic state.

6
Previous Notions
 PRINCIPAL STRESSES
 Regardless of the state of stress, it is always possible to choose a special
set of axes (principal axes of stress or principal stress directions) so
that the shear stress components vanish when the stress components are
referred to this system.
 The three planes perpendicular to the principle axes are the principal
planes.
 The normal stress components in the principal planes are the principal
stresses. σ 33
x3 σ 31 σ 32 x3
σ 13 x3′
σ 1 0 0  σ 23
σ 11 σ 12 σ 21 σ x1′ σ3
[σ ] =  0 σ 2 0  22
σ1
 0 0 σ 3  σ2
x1 x1
x2 x2
x2′
7
Previous Notions
 PRINCIPAL STRESSES
 The Cauchy stress tensor is a symmetric 2nd order tensor so it will diagonalize
in an orthonormal basis and its eigenvalues are real numbers.
 Computing the eigenvalues λ and the corresponding eigenvectors v:
σ ⋅ v λv
= [σ −=
λ 1] ⋅ v 0 σ 11 − λ σ 12 σ 13
not
det [ σ − λ 1] = =
σ − λ 1 σ 12 σ 22 − λ =
σ 23 0
σ 13 σ 23 σ 33 − λ
INVARIANTS
σ 33
λ 3 − I1λ 2 − I 2 λ − I 3 =
characteristic x3 σ 31 σ 32 x3
0 equation σ 13 x3′
σ 23
λ1 ≡ σ 1 σ 11 σ 12 σ 21 σ x1′ σ3
σ1
22
λ2 ≡ σ 2
λ3 ≡ σ 3 σ2
x1 x1
x2 x2
x2′
8
Previous Notions
 STRESS INVARIANTS
 Principal stresses are invariants of the stress state.
 They are invariant w.r.t. rotation of the coordinate axes to which the
stresses are referred.
 The principal stresses are combined to form the stress invariants I :
I1 = Tr ( σ ) = σ ii = σ 1 + σ 2 + σ 3 REMARK
I 2 =( σ : σ − I12 ) =
1
− (σ 1σ 2 + σ 1σ 3 + σ 2σ 3 ) The I invariants are obtained
2 from the characteristic equation
I 3 = det ( σ ) of the eigenvalue problem.
 These invariants are combined, in turn, to obtain the invariants J:
J=
1 I=
1 σ ii REMARK
The J invariants can be
J2 =
2
(
1 2
I1 + 2 I 2 ) = σ ijσ ji = ( σ : σ )
1
2
1
2 expressed the unified form:
1 = Ji
1
( )
Tr σ i i ∈ {1, 2,3}
J=
3
3
( I1 + 3I1 I 2 + 3I3=) 3 Tr ( σ ⋅ σ ⋅ σ=) 3 σ ijσ jkσ ki
1 3 1
i
9
Previous Notions
 SPHERICAL AND DEVIATORIC PARTS OF THE STRESS TENSOR
Given the Cauchy stress tensor σ and its principal stresses, the following is
defined:
 Mean stress
1 1 1
σ m= Tr (σ )= σ ii = (σ 1 + σ 2 + σ 3 )
3 3 3
REMARK
 Mean pressure
In a hydrostatic state of stress, the
1
p= −σ m =− (σ 1 + σ 2 + σ 3 ) stress tensor is isotropic and, thus,
3 its components are the same in
any Cartesian coordinate system.
 A spherical or hydrostatic
As a consequence, any direction
state of stress: σ 0 0 is a principal direction and the
σ=
1 σ=
2 σ3 σ ≡  0 
σ 0 = σ 1 stress state (traction vector) is the
 0 0 σ  same in any plane.

10
Previous Notions
 SPHERICAL AND DEVIATORIC PARTS OF THE STRESS TENSOR
The Cauchy stress tensor σ can be split into: σ = σ sph + σ ′
 The spherical stress tensor:
 Also named mean hydrostatic stress tensor or volumetric stress tensor or
mean normal stress tensor.
 Is an isotropic tensor and defines a hydrostatic state of stress.
 Tends to change the volume of the stressed body
1 1
σ=
sph : σ=m 1 Tr ( σ=) 1 σ ii 1
3 3
 The stress deviatoric tensor:
 Is an indicator of how far from a hydrostatic state of stress the state is.
 Tends to distort the volume of the stressed body
σ ′ dev
= = σ σ − σm 1

11
Previous Notions
 STRESS INVARIANTS OF THE STRESS DEVIATORIC TENSOR
 The stress invariants of the stress deviatoric tensor:
=I1′ Tr =( σ′ ) 0
=I 2′
1
2
(
σ′ : σ′ − I12 )
′ = (σ ij′σ ′jkσ ki′ )
1
I 3′ = det ( σ′ ) = σ 11′ σ 22
′ σ 33
′ + 2σ 12′ σ 23
′ σ 13′ − σ 12′2σ 33
′ − σ 23
′2σ 11′ − σ 13′2σ 22
3

 These correspond exactly with the invariants J of the same stress


deviator tensor:
′ I=
J=
1
′ 0
1

J 2′ =
1 2
2
( )
I1′ + 2 I 2′ =
1
I 2′ = ( σ′ : σ′ )
2

J 3′ =
3
( )
I1′ + 3I1′I 2′ + 3I 3′ = I 3′ = Tr ( σ′ ⋅ σ′ ⋅ σ′ ) = (σ ij′σ ′jkσ ki′ )
1 3 1
3
1
3

12
Previous Notions
 EFFECTIVE STRESS
 The effective stress or equivalent uniaxial stress σ is the scalar:

3 3
σ= 3J=
'
2 σ′ij σ′ij= σ´: σ´
2 2

 It is an invariant value which measures the “intensity” of a 3D stress state


in a terms of an (equivalent) 1D tensile stress state.
 It should be “consistent”: when applied to a real 1D tensile stress, should
return the intensity of this stress.

13
Example
Calculate the value of the equivalent uniaxial stress for an uniaxial state of
stress defined by:
E, ν
y
σ u σ 0x0
σx
σ ≡  0 0 0 
σu σu
 0 0 0 
x

14
σ u 0 0
σ ≡  0 0 0 
Example - Solution  0 0 0 

1 σu σ u 
Mean stress: =
σm =
Tr (σ ) 0 0
3 3  
σ m 0 0 3 
σ
Spherical and deviatoric parts σ sph ≡  0 σ m 0  =0 u
0
 3 
 0 0 σ m   
of the stress tensor: 0 σ u 
0
 3 
2 
 3 σu 0 0 
σ u − σ m 0 0   
≡  0 0  =  0 0 
1
σ′ =−
σ σ sph −σ m − σu
 3 
 0 0 −σ m   
 0 1 
0 − σu
 3 
3 3 2 4 1 1 32
σ= σ′ij σ′ij= σu ( + + =) σu σ = σu
2 2 9 9 9 23

15
8.2 Principal Stress Space
Ch.8. Plasticity

16
Principal Stress Space
 The principal stress space or Haigh–Westergaard stress space is
the space defined by a system of Cartesian axes where the three
spatial axes represent the three principal stresses for a body
subject to stress:
σ1 ≥ σ 2 ≥ σ 3

17
Octahedral plane
 Any of the planes perpendicular to the hydrostatic stress axis is a
octahedral plane. 1
1  
 Its unit normal is n = 1 .
3 
1 σ1 ≥ σ 2 ≥ σ 3

18
Normal and Shear Octahedral Stresses
 Consider the principal stress space:
 The normal octahedral stress is defined as:

1/ 3 
 
3σ oct = OA = OP ⋅ n= [σ 1 , σ 2 , σ 3 ] 1/ 3 =
 
1/ 3 
3
= (σ 1 + σ 2 + σ=
3) 3 σm
3

I1
σ=
oct σ=
m
3
19
Normal and Shear Octahedral Stresses
 Consider the principal stress space:
 The shear or tangential octahedral
stress is defined as:
3τ oct = AP

 Where the AP is calculated from:

= AP = OP − OA = (σ 12 + σ 22 + σ 32 ) −
2 2 2
3τ 2
oct
1
− (σ 1 + σ 2 + σ 3 ) =
2
2 J 2'
3 Alternative forms of τ oct :
1/2
1  2 1 2
τ= σ + σ 2
+ σ 2
− ( σ + σ + σ )
3  
oct 1 2 3 1 2 3
3
2
[ J2 ] ( ) ( ) ( )
1  2 1/2
′ 
τ oct
2 2
=
12
τ oct
= σ − σ + σ − σ + σ − σ
3 3  
1 2 2 3 1 3
3
20
Normal and Shear Octahedral Stresses
 In a pure spherical stress state:
1
σ= σ 1 → σm= 3σ → σ esf= σ=
1 σ
3
σ′ =σ − σ esf = 0 J 2′ = 0
A pure spherical stress state is
τ oct = 0 located on the hydrostatic stress axis.

 In a pure deviator stress state:

σ = σ′ σ m Tr
= = (σ′) 0
(σ ) Tr= σ oct = 0

A pure deviator stress state is located on the octahedral


plane containing the origin of the principal stress space

21
Stress Invariants
 Any point in space is unambiguously defined by the three
invariants:
 The first stress invariant I1 characterizes the distance from the origin to
the octahedral plane containing the point.

 The second deviator stress invariant J 2′ characterizes the radius of the


cylinder containing the point and with the hydrostatic stress axis as axis.

 The third deviator stress invariant J 3′


characterizes the position of the point on
the circle obtained from the intersection
of the octahedral plane and the
cylinder. It defines an angle θ ( J 3′ ) .

22
Projection on the Octahedral Plane
 The projection of the principal stress space on the octahedral
plane results in the division of the plane into six “sectors”:
 These are characterized by the different principal stress orders.

Election of a criterion, FEASIBLE


e.g.: σ 1 ≥ σ 2 ≥ σ 3 WORK SPACE

23
8.4 Phenomenological Behaviour
Ch.8. Plasticity

38
Notion of Plastic Strain
PLASTIC STRAIN

ε= ε e + ε p

elastic limit: σ e

LINEAR ELASTIC
BEHAVIOUR
σ = Eεe

39
Bauschinger Effect
 Also known as kinematic hardening.

σf
σ e + Kε

K
σe

−σ e + K ε

K
−σ e

41
Elastoplastic Behaviour
 Considering the phenomenological behaviour observed,
elastoplastic materials are characterized by:

 Lack of unicity in the stress-strain relationship.


 The stress value depends on the actual strain and the previous loading
history.

 A nonlinear stress-strain relationship.


 There may be certain phases in the deformation process with
incremental linearity.

 The existence of permanent (or plastic) strain during a loading /


unloading cycle.

42
8.5 1D Incremental Plasticity Theory
Ch.8. Plasticity

43
Introduction
 The incremental plasticity theory is a mathematical model used
to represent the evolution of the stress-strain curve in an
elastoplastic material.
 Developed for 1D but it can be generalized for 3D problems.

REMARK
This theory is
developed under
the hypothesis of
infinitesimal strains.

44
Additive Decomposition
of Strain
Total strain can be split into an elastic (recoverable) part, ε ,
e

and an inelastic (unrecoverable) one, ε :
p

σ
ε= ε + ε
e p
where ε = e

E elastic modulus or
Young modulus

 Also,

dε dε + dε
= e p
where dε =e

45
Hardening Variable
 The hardening variable, α , is defined as: REMARK
The sign ( • ) function is:
dα = sign (σ ) d ε p

Such that dα ≥ 0 and α ε p =0 = 0.

 Note that α is always positive and:


dα d=
= α sign (σ ) d ε p dα = d ε p
 
=1
Then, for a monotonously increasing plastic strain process, both
variables coincide:
εp εp
dε ≥ 0
p
= α ∫= dε p
∫ d ε p
= ε p
0 0

46
Yield Stress and Hardening Law
 Stress value, σ f , threshold for the material exhibiting plastic
behaviour after elastic unloading + elastic loading
 It is considered a material property.
 For ε p= α= 0 σ f= σ e HARDENING LAW
σ=
f σ e + H ′α
σ = σ f (α )

dσ f = H ′dα
σ ≤ σ f (α )

 H ′ is the hardening modulus

47
Yield Function
 The yield function, F (σ , α ) , characterizes the state of the material:

F (σ , α ) ≡ σ − σ f (α )

Space of
F (σ , α ) < 0 F (σ , α ) = 0 admissible
stresses
ELASTIC STATE ELASTO-PLASTIC STATE

{
σ ∈ R F (σ , α ) < 0
Eσ := } {
∂Eσ := σ ∈ R F (σ , α ) = 0 }
ELASTIC DOMAIN YIELD SURFACE
INITIAL ELASTIC
DOMAIN: {
Eσ0 := σ ∈ R F (σ , 0 ) ≡ σ − σ e < 0 }
48
Space of Admissible Stresses
 Any admissible stress state must belong to the space of
admissible stresses, Eσ (postulate):
Eσ = Eσ  ∂Eσ =
{σ ∈ R F (σ , α ) ≤ 0}
=

F (σ , α ) ≡ σ − σ f (α )

Space of
admissible
stresses
REMARK

Eσ ≡  −σ f (α ), σ f (α ) 

49
Constitutive Equation
 The following situations are defined:
 ELASTIC REGIME

σ ∈ Eσ dσ = E d ε

 ELASTOPLASTIC REGIME – UNLOADING


σ ∈ ∂Eσ
dσ = E d ε
dF (σ , α ) < 0

 ELASTOPLASTIC REGIME – PLASTIC LOADING


REMARK
σ ∈ ∂Eσ The situation
σ ∈ ∂Eσ
dσ = E d ε
ep
dF (σ , α ) > 0
dF (σ , α ) = 0
is not possible because, by
Elastoplastic
definition, on the yield
tangent modulus
surface F (σ , α ) = 0 .
50
Elastoplastic Tangent Modulus
 Consider the elastoplastic regime in plastic loading,
σ ∈ ∂Eσ
F (σ , α ) =
≡ σ − σ f (α ) 0 dF (σ , α ) 0
=
∂σ
dF (σ , α ) = dσ − σ ′f (α ) dα = 0
∂σ
 
= sign (σ ) =H ′
1
dα = sign (σ ) dσ
H′
 Since the hardening variable is defined as:
dα = sign (σ ) d ε p
dσ f = H ′dα
1
=dε p
dσ for σ ∈ ∂Eσ HARDENING LAW
H ′
51
Elastoplastic Tangent Modulus
1
Elastic strain dε e = dσ
E
1
Plastic strain dε =
p

H′
Additive strain decomposition :
1 1
d ε =d ε + d ε =( + )dσ
e p

E H′
1 EH ′
=dσ = dε dε
1 1
+ E + H′

E H′ E ep
ELASTOPLASTIC EH ′
dσ = E d ε
ep E =
ep
TANGENT MODULUS E + H′
52
Uniaxial Stress-Strain Curve
 Following the constitutive equation defined

ELASTOPLASTIC
REGIME
dσ = E ep d ε

ELASTIC
REGIME
dσ = E d ε

REMARK
Plastic strain is generated only
during the plastic loading process.

53
Role of the Hardening Modulus
 The value of the hardening modulus, H ′, determines the following
situations:
EH ′
E =
ep

E + H′

H′ > 0
Linear elasticity Plasticity with
strain hardening H′ < 0
Plasticity with
strain softening

Perfect plasticity

55
Plasticity in Real Materials
 In real materials, the stress-strain curve shows a combination of
the three types of hardening modulus.

H′ < 0
H′ = 0 H′ > 0

56
8.6 3D Incremental Theory
Ch.8. Plasticity

57
Introduction
 The 1D incremental plasticity theory can be generalized to a
multiaxial stress state in 3D.

The same concepts are used:


 Additive decomposition of strain
 Hardening variable
 Yield function

Plus, additional ones are added:


 Loading - unloading conditions
 Consistency conditions

58
ε= ε e + ε p
Additive Decomposition 
1D →  e σ
ε = E
of Strain
Total strain can be split into an elastic (recoverable) part, ε ,
e

and an inelastic (unrecoverable) one, ε :
p

ε= ε e + ε p where ε e = C −1 : σ

constitutive elastic
(constant) tensor
 Also,

dε dε + dε
= e p
where d ε e = C −1 : dσ

59
1D → d ε p =λ sign (σ )
Hardening Variable =d α

 The hardening variable, α = f (σ , ε p ) , is a scalar:

dα = λ with α ∈ [ 0, ∞ )

Where λ is known as the plastic multiplier.

 The flow rule is defined as:


∂G (σ , α )
dε = λ
p

∂σ

Where G (σ , α ) is the plastic potential function

60
Yield Function 1D → F (σ , α ) ≡ σ − σ f (α )

 The yield function, F (σ , α ) , is a scalar defined as:


Equivalent
F (σ , α ) ≡ φ (σ ) − σ f (α ) uniaxial stress

Yield stress

F (σ , α ) < 0 F (σ , α ) = 0
ELASTIC STATE ELASTOPLASTIC STATE

Eσ : {σ }
F (σ , α ) < 0 =
∂Eσ : {σ (σ , α ) 0
F= }
ELASTIC DOMAIN YIELD SURFACE
Space of
INITIAL ELASTIC
=
DOMAIN: E 0
σ : {σ F (σ , 0 ) < 0 } admissible =
stresses
Eσ Eσ  ∂Eσ

61
Loading-Unloading Conditions and
Consistency Condition
 Loading/unloading conditions (also known as Karush-Kuhn-
Tucker conditions):
λ≥0 ; F ( σ, α ) ≤ 0 ; λ F ( σ, α ) =
0

 Consistency conditions:
For F ( σ, α ) =
0 → λ dF ( σ, α ) =
0

∂G (σ , α ) ELASTOPLASTIC
F=0; dF < 0 λ= 0; d ε = λ p
=0 ELASTIC UNLOADING
∂σ
 ∂G (σ , α ) ELASTOPLASTIC
=  λ =
0; d ε p
λ = 0 NEUTRAL LOADING
∂σ
=F 0;=dF 0 
λ > 0; d ε= ∂G (σ , α ) ELASTOPLASTIC
p
λ ≠0
 ∂σ LOADING
=F 0; dF > 0 IMPOSSIBLE
62
Constitutive Equation
 The following situations are defined:
 ELASTIC REGIME ( F < 0)

σ ∈ Eσ dσ = C : d ε

 UNLOADING( F 0 and dF (σ , α ) < 0)


ELASTOPLASTIC REGIME – ELASTIC=
σ ∈ ∂Eσ
dσ = C : d ε
dF (σ , α ) < 0

 =
ELASTOPLASTIC REGIME – PLASTIC and dF (σ , α ) 0)
LOADING ( F 0=
σ ∈ ∂Eσ
dσ = C ep : d ε
dF (σ , α ) = 0
ELASTOPLASTIC
CONSTITUTIVE TENSOR

63 MMC - ETSECCPB - UPC 26/04/2017


Elastoplastic Constitutive Tensor
 The elastoplastic constitutive tensor is written as:
REMARK
∂G ∂F When the plastic potential function

C: :C
) C − ∂σ∂F ∂σ ∂G
C ep ( σ, α = and the yield function coincide, it is
said that there is associated flow:
H′+ :C:
∂σ ∂σ G (σ , α ) = F (σ , α )

∂G ∂F
Cijpq C rskl
∂σ pq ∂σ rs
Cijkl
ep
=
Cijkl − i, j , k , l , p, q, r , s ∈ {1, 2,3}
∂F ∂G
H′+ C pqrs
∂σ pq ∂σ rs

64
8.7 Failure Criteria: Yield Surfaces
Ch.8. Plasticity

65
Introduction
 The initial yield surface, ∂Eσ0 , is the external boundary of the initial
elastic domain Eσ for the virgin material
0

 The state of stress inside the yield surface is elastic for the virgin material.
 When in a deformation process, the stress state reaches the yield surface, the
virgin material looses elasticity for the first time: this is considered as a failure
criterion for design. Subsequent stages in the deformation process are not
considered.

66 MMC - ETSECCPB - UPC 26/04/2017


Yield (Failure) Criteria
 The yield surface is usually expressed in terms of the following
invariants to make it independent of the reference system (in the
principal stress space):
F ( σ ) ≡ F ( I1 , J 2′ , J 3′ ) − σ e =
0
  with σ 1 ≥ σ 2 ≥ σ 3
φ( σ )
REMARK
 Where:
Due to the adopted principal stress
I1 = Tr ( σ ) = σ ii = σ 1 + σ 2 + σ 3 criteria, the definition of yield
surface only affects the first sector
J 2′ =
2
(
1 2
)
I1′ + 2 I 2′ =
1
I 2′ = ( σ′ : σ′ )
2
of the principal stress space.

J 3′ =
3
(
1 3
) 1
3
1
(
I1′ + 3I1′I 2′ + 3I 3′ = I 3′ = Tr ( σ′ ⋅ σ′ ⋅ σ′ ) = σ ij′σ ′jkσ ki′
3
)
 The elastoplastic behavior will be isotropic.

67
F ( σ ) ≡ φ( σ ) − σ e = 0
Von Mises Criterion
 The yield surface is defined as: REMARK
The Von Mises criterion
F (σ ) ≡ σ ( ) − σ e =
0 depends solely on the
second deviator stress
 Where σ (σ ) = 3J 2′ is the effective stress. invariant.
(often termed the Von-Mises stress)
2
The shear octahedral stress is, by definition, τ oct = [ J2 ] .

12

3
Thus, the effective stress is rewritten:
3 3
σ (σ )
3
[ J 2′ ] = τ oct = 3 τ oct
= τ oct
12

2 2 2

3
 And the yield surface is given by: F (σ ) ≡ τ oct − σ e =
0
2

68
3
Von Mises Criterion F (σ ) ≡
2
τ oct − σ e =
0

 The octahedral stresses characterizes the radius of the cylinder


containing the point and with the hydrostatic stress axis as axis.
2 2
τ oct = σ e → 3τ oct = σe
3 3

F (σ) ≡ F ( J 2′ ) − σ e
REMARK
The Von Mises Criterion is adequate for metals, where
hydrostatic stress states have an elastic behavior and
failure is typically due to deviatoric stress components.

69
Example
Consider a beam under a composed flexure state such that for a beam section
the stress state takes the form,

σ x τ xy 0 
σ
[σ ] = τ xy 0 x 0
σx
 0 0 0 

Obtain the expression for Von Mises criterion.

70
Example - Solution
The mean stress is: 1 σx
=σm =Tr ( σ )
3 3
The deviator part of the stress tensor is:

σ x − σ m τ xy 0   23 σ x τ xy 0 
 
σ′ =σ − σ esf ≡  τ xy −σ m 0  =  τ xy − 13 σ x 0 
 0 0 −σ m   0 0 − 13 σ x 

The second deviator stress invariant is given by,


1 14 1 1  1
J 2′ = σ ′ : σ ′ =  σ x2 + σ x2 + σ x2 + τ xy2 + τ xy2  = σ x2 + τ xy2
2 29 9 9  3

71
Example - Solution
The uniaxial effective stress is:
σ (=
σ) J 2′
3= σ x2 + 3τ xy2

Finally, the Von Mises yield surface is given by the expression:

F (σ ) ≡ 3 J 2′ − σ e =
0 σ x2 + 3τ xy2 σe
=

co
(comparison stress)

(Criterion in design codes for metal beams)

72
F ( σ ) ≡ φ( σ ) − σ e = 0
Tresca Criterion
 Also known as the maximum shear stress criterion, it establishes
that the elastic domain ends when:
σ1 − σ 3 σ e
= τ max = F (σ ) ≡ (σ 1 − σ 3 ) − σ e =
0
2 2
Plane parallel to axis σ 2

 It can be written univocally in terms of invariants J 2′ and J 3′ :


F (σ ) ≡ (σ 1 − σ 3 ) − σ e ≡ F ( J 2′ , J 3′ ) − σ e

73
Tresca Criterion F (σ ) ≡ (σ 1 − σ 3 ) − σ e =
0

F (σ) ≡ F ( J 2′ , J 3′ ) − σ e =
0

REMARK
The Tresca yield surface is appropriate for metals, which have an elastic behavior under
hydrostatic stress states and basically have the same traction/compression behavior.

74
Example
Obtain the expression of the Tresca criterion for an uniaxial state of stress
defined by:
E, G
y
σ u 0 0
σ ≡  0
σ0x 0 σx
 σu σu
 0 0 0 
x

76
σ u 0 0
σ ≡  0 0 0 
Example - Solution  0 0 0 

Consider:

σ1 = σ u
σu ≥ 0 F (σ ) = (σ 1 − σ 3 ) − σ e = σ u − σ e = σ u − σ e
σ3 = 0 
σu

σ1 = 0
σu < 0 F (σ ) =
(σ 1 − σ 3 ) − σ e =
−σ u − σ e =σu −σe
σ3 = σu 
σu

The Tresca criterion is expressed as:


Note that it coincides with
F (σ ) ≡ σ − σ e =
0 σu = σe the Von Mises criterion for an
uniaxial state of stress.

77
Example
Consider a beam under a composed flexure state such that for a beam section
the stress state takes the form,

σ x τ xy 0 
σ
[σ ] = τ xy 0 x 0
σx
 0 0 0 

Obtain the expression for Tresca yield surface.

78
Example - Solution
The principal stresses are:
1 1 1 1
σ 1 = σ x + σ x2 + τ xy2 , σ 3 = σ x − σ x2 + τ xy2
2 4 2 4

Taking the definition of the Tresca yield surface,

F (σ ) ≡ (σ 1 − σ 3 ) − σ e =
0
1 1 2 2  1 1 2 2 
σ e = σ 1 − σ 3 =  σ x + σ x + τ xy  −  σ x − σ x + τ xy 
2 4  2 4 
σ x2 + 4τ xy2 σe
=

co
(comparison stress)
79
Mohr-Coulomb Criterion
 It is a generalization of the Tresca criterion, by including the
influence of the first stress invariant.
 In the Mohr circle’s plane, the Mohr-Coulomb yield function takes
the form, cohesion internal friction angle
τ = c − σ tan φ
REMARK
The yield line cuts the
normal stress axis at a
positive value, limiting the
materials tensile strength.

80
Mohr-Coulomb Criterion
 Consider the stress state for which the yield point is reached:

τ A = R cos φ
σ1 + σ 3
σA
= + R sin φ
2

σ1 − σ 3 σ + σ 3 σ1 − σ 3 
τ A + σ A tg φ − c 0
= cos φ +  1 + = sin φ  tg φ − c 0
2  2 2 
(σ 1 − σ 3 ) + (σ 1 + σ 3 ) sin φ − 2c cos φ = 0 REMARK
For φ 0=
= and c σ e / 2 ,
F (σ ) ≡ (σ 1 − σ 3 ) + (σ 1 + σ 3 ) sin φ − 2c cos φ =
0 the Tresca criterion is
recovered.

81
Mohr-Coulomb Criterion

F (σ) ≡ F ( I1 , J 2′ , J 3′ ) =
0

REMARK
The Mohr-Coulomb yield surface is appropriate for frictional cohesive
materials, such as concrete, soils or rocks which have considerably
different tensile and compressive values for the uniaxial elastic limit.

82
Drucker-Prager Criterion
 It is a generalization of the Von Mises criterion, by including the
influence of the first stress invariant.
 The yield surface is given by the expression: REMARK
For φ 0=
= and c σ e / 2 ,
F (σ ) ≡ 3ασ m + [ J 2′ ] − β =
1/2
0
the Von Mises criterion
is recovered.
Where: 
2sin φ 6c cos φ σ 1 + σ 2 + σ 3 I1
=α = ; β =; σm =
3 ( 3 − sin φ ) 3 ( 3 − sin φ ) 3 3

 It can be rewritten as:


3
F (σ ) ≡ α I1 + [ J 2′ ] −=
β 3ασ oct + β  ( I1′, J 2′ )
τ oct −=
1/2

83
Drucker-Prager Criterion

F (σ) ≡ F ( I1 , J 2′ )
REMARK
The Drucker-Prager yield surface, like the Mohr-Coulomb one, is appropriate for
frictional cohesive materials, such as concrete, soils or rocks which have
considerably different tensile and compressive values for the uniaxial elastic limit.

84
Chapter 8
Plasticity

rs
ee
s gin
8.1 Introduction

t d le En

r
The elastoplastic models (constitutive equations) are used in continuum mechan-

ba
ge ro or
eS m
ics to represent the mechanical behavior of materials whose behavior, once cer-

ci
f
tain limits in the values of the stresses (or strains) are exceeded, is no longer rep-

ra
C d P cs
resentable by means of simpler models such as the elastic ones. In this chapter,
b
a
i
these models will be studied considering, in all cases, that strains are infinitesi-
an an n

mal.
y ha

Broadly speaking, plasticity introduces two important modifications with re-


le
liv or ec

spect to the lineal elasticity seen in chapters 6 and 7:


M

.A

1) The loss of linearity: stresses cease to be proportional to strains.


m

2) The concept of permanent or plastic strain: a portion of the strain generated


d

during the loading process is not recovered during the unloading process.
uu
e
X Th

er
tin

8.2 Previous Notions


on

.O

The concepts in this section are a review of those already studied in Sec-
C

tions 4.4.4 to 4.4.7 of Chapter 4.


©

8.2.1 Stress Invariants


Consider the Cauchy stress tensor σ and its matrix of components in a base
associated with the Cartesian axes {x, y, z} (see Figure 8.1),
⎡ ⎤
σx τxy τxz
⎢ ⎥
σ ]xyz = ⎣ τxy σy τyz ⎦ .
[σ (8.1)
τxz τyz σz

369
370 C HAPTER 8. P LASTICITY

diagonalization

rs
ee
s gin
Figure 8.1: Diagonalization of the stress tensor.

t d le En
Since σ is a symmetrical second-order tensor, it will diagonalize in an orthonor-

r
ba
ge ro or
eS m
mal base and all its eigenvalues will be real numbers. Then, consider a system

ci
of Cartesian axes {x , y , z } associated with a base in which σ diagonalizes. Its
f

ra
C d P cs
matrix of components in this base is
b
a
i
⎡ ⎤
an an n

σ1 0 0
y ha

⎢ ⎥
σ ]x  y  z  = ⎣ 0 σ 2 0 ⎦ ,

le
(8.2)
liv or ec

0 0 σ3
M

.A

where σ1 ≥ σ2 ≥ σ3 , denoted as principal stresses, are the eigenvectors of σ and


m

the directions associated with the axes {x , y , z } are named principal directions
uu
e

(see Figure 8.1).


X Th

er

To obtain the stresses and the principal directions of σ , the corresponding


tin

eigenvalue and eigenvector problem must be solved:


on

.O

Find λ and v such that σ · v = λ v =⇒ σ − λ 1) · v = 0 ,


(σ (8.3)
C

where λ corresponds to the eigenvalues and v to the eigenvectors. The necessary


and sufficient condition for (8.3) to have a solution is
 
σ − λ 1) = σ − λ 1 = 0 ,
det (σ (8.4)
which, in component form, results in
 
 σx − λ τxy τxz 

 
 τxy σy − λ τyz  = 0 . (8.5)
 
 τxz τyz σz − λ 

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Previous Notions 371

The algebraic development of (8.5), named characteristic equation, corre-


sponds to a third-degree polynomial equation in λ , that can be written as

λ 3 − I1 λ 2 − I2 λ − I3 = 0 , (8.6)

where the coefficients I1 (σi j ), I2 (σi j ) and I3 (σi j ) are certain functions of the
components σi j of the tensor σ expressed in the coordinate system {x, y, z}. Yet,
the solutions to (8.6), which will be a function of its coefficients (I1 , I2 , I3 ), are
the principal stresses that, on the other hand, are independent of the system of
axes chosen to express σ . Consequently, said coefficients must be invariant with

rs
respect to any change of base. Therefore, the coefficients I1 , I2 and I3 are denoted

ee
as I stress invariants or fundamental stress invariants and their expression (re-

s gin
sulting from the computation of (8.5)) is

t d le En

⎪ I = Tr (σσ ) = σii = σ1 + σ2 + σ3
⎨ 1
1 

r
I stress

ba
I2 = σ : σ − I12 = − (σ1 σ2 + σ1 σ3 + σ2 σ3 )

ge ro or
(8.7)

eS m
invariants ⎪
⎪ 2

ci

σ ) = σ1 σ2 σ3
I3 = det (σ f

ra
C d P cs
b
a
i
an an n
y ha

Obviously, any scalar function of the stress invariants will also be an invariant
le
liv or ec

and, thus, new invariants can be defined based on the I stress invariants given
in (8.7). In particular, the so-called J stress invariants are defined as
M

.A


m



d


⎪ σ)
uu


⎪ J1 = I1 = σii = Tr (σ
e



X Th


er


tin


⎪ 1 2  1

⎪ J2 = I1 + 2I2 = σi j σ ji =


on


.O

2 2
J stress 1 1 (8.8)
= (σ σ : σ ) = Tr (σ σ ·σ)
C

invariants ⎪
⎪ 2 2

©




⎪ 1 3  1

⎪ = + + = σi j σ jk σki =

⎪ J3 I1 3I1 I2 3I3

⎪ 3 3

⎪ 1

⎩ = Tr (σ σ ·σ ·σ)
3

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
372 C HAPTER 8. P LASTICITY

Remark 8.1. Note that


I1 = 0 =⇒ Ji = Ii i ∈ {1, 2, 3} .
Also, the invariants Ji , i ∈ {1, 2, 3} can be expressed in a unified
and compact form by means of
1 
Ji = Tr σ i i ∈ {1, 2, 3} .
i

rs
ee
8.2.2 Spherical and Deviatoric Components of the Stress Tensor

s gin
Given the stress tensor σ , the mean stress σm is defined as

t d le En
I1 1 1 1
σm = σ ) = σii = (σ1 + σ2 + σ3 )
= Tr (σ (8.9)

r
ba
3 3 3 3

ge ro or
eS m
ci
f

ra
and the mean pressure p̄ as
C d P cs
b
a
p̄ = −σm . (8.10)
i
an an n
y ha

The Cauchy stress tensor can be decomposed into a spherical part (or com-
le
ponent), σ sph , and a deviatoric one, σ  ,
liv or ec
M

.A

σ = σ sph + σ  , (8.11)
m

where the spherical part of the stress tensor is defined as


uu
e
X Th

er
tin

de f 1
σ sph : = Tr (σ σ ) 1 = σm 1
on

.O

⎡3 ⎤
C

σm 0 0 (8.12)
not ⎢ ⎥
©

σ sph ≡ ⎣ 0 σm 0 ⎦
0 0 σm

and, from (8.11) and (8.12), the deviatoric part is given by

⎡ ⎤
σx − σm τxy τxz
not ⎢ ⎥
σ  = σ − σ sph ≡ ⎣ τxy σy − σm τyz ⎦ . (8.13)
τxz τyz σz − σm

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Previous Notions 373

Finally, the I and J invariants of the deviatoric tensor σ  , named I  and J  invari-
ants, respectively, are derived from (8.7), (8.8), (8.9) and (8.13).




⎪ J1 = I1 = 0




J  stress 1  1
invariants ⎪ J2 = I2 = (σ σ : σ  ) = σij σ ji (8.14)

⎪ 2 2

⎪  

⎪ 1
⎩ J3 = I3 = σ σ σ
3 i j jk ki

rs
ee
s gin
Remark 8.2. It is easily proven that the principal directions of σ co-

t d le En
incide with those of σ  , that is, that both tensors diagonalize in the

r
same base. In effect, working in the base associated with the princi-

ba
ge ro or
pal directions of σ , i.e., the base in which σ diagonalizes, and, given

eS m
ci
that σ sph is a hydrostatic tensor and, thus, is diagonal in any base,
f

ra
then σ  also diagonalizes in the same base (see Figure 8.2).
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 8.2: Diagonalization of the spherical and deviatoric parts of the stress tensor.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
374 C HAPTER 8. P LASTICITY

Remark 8.3. The effective stress or equivalent uniaxial stress σ̄ is the


scalar
  
 3   3 
σ̄ = 3 J2 = σi j σ ji = σ : σ .
2 2
The name of equivalent uniaxial stress is justified because its value
for an uniaxial stress state coincides with said uniaxial stress (see
Example 8.1).

rs
ee
s gin
Example 8.1 – Compute the value of the equivalent uniaxial stress (or effec-
tive stress) σ̄ for an uniaxial stress state defined by

t d le En
⎡ ⎤

r
σu 0 0

ba
ge ro or
not ⎢ ⎥

eS m
σ ≡⎣ 0 0 0 ⎦ .

ci
f

ra
C d P cs
0 0 0
b
a
i
an an n
y ha

Solution
le
liv or ec

The mean stress is


σ
M

.A

1
σ) = u .
Tr (σ σm =
3 3
m

Then, the spherical component of the stress tensor is


uu
e

⎡ ⎤
X Th

σu
er

⎡ ⎤
tin

0 0
σm 0 0 ⎢ 3 ⎥
not ⎢ ⎥ ⎢ σu ⎥
on

σ sph ≡ ⎣ 0 σm 0 ⎦ = ⎢ ⎥
.O

⎢ 0 3 0 ⎥
⎣ σu ⎦
C

0 0 σm
©

0 0
3
and the deviatoric component results in
⎡ ⎤
⎡ ⎤ 2
σu 0 0
σu − σm 0 0 ⎢3 ⎥
⎢ ⎥ ⎢⎢ 1 ⎥

0 ⎥
not
σ = σ − σ sph ≡ ⎣ 0 −σm 0 ⎦ = ⎢ 0 − σu ⎥.
⎣ 3
1 ⎦
0 0 −σm
0 0 − σu
3

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Principal Stress Space 375

Finally, the equivalent uniaxial stress (or effective stress) is obtained,


    
3   3 2 4 1 1 32
σ̄ = σi j σ ji = σu + + = |σu | = |σu | =⇒
2 2 9 9 9 23

σ̄ = |σu | .

rs
8.3 Principal Stress Space

ee
s gin
Consider a system of Cartesian axes in R3 {x ≡ σ1 , y ≡ σ2 , z ≡ σ3 } such that
each stress state, characterized by the values of the three principal stresses

t d le En
σ1 ≥ σ2 ≥ σ3 , corresponds to a point in this space, which is known as the prin-
cipal stress space1 (see Figure 8.3).

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
Definition 8.1. The hydrostatic stress axis is the locus of points in
b
a
the principal stress space that verify the condition σ1 = σ2 = σ3 (see
i
an an n

Figure 8.3). The points located on the hydrostatic stress axis repre-
y ha

sent hydrostatic states of stress (see Chapter 4, Section 4.4.5).


le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin

hydrostatic stress axis


on

.O

(σ1 = σ2 = σ3 )
= bisector of the 1st octant
C

Figure 8.3: The principal stress space.

1 The principal stress space is also known as the Haigh-Westergaard stress space.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
376 C HAPTER 8. P LASTICITY

hydrostatic stress axis


(σ1 = σ2 = σ3 )

rs
ee
Figure 8.4: The hydrostatic stress axis and the octahedral plane.

s gin
t d le En
Definition 8.2. The octahedral plane Π is any of the planes that

r
ba
ge ro or
eS m
are perpendicular to the hydrostatic stress axis (see Figure 8.4). The

ci
equation of an octahedral plane is f

ra
C d P cs
σ1 + σ2 + σ3 = const.
b
a
i
an an n

and the unit normal vector of said plane is


y ha

not 1
n ≡ √ [1, 1, 1]T .
le
liv or ec

3
M

.A
m

d
uu
e

8.3.1 Normal and Shear Octahedral Stresses


X Th

er
tin

Consider P is a point in the principal stress space with coordinates (σ1 , σ2 , σ3 ).


The position vector of this point is defined as OP ≡= [σ1 , σ2 , σ3 ]T (see Fig-
not
on

.O

ure 8.5). Now, the octahedral plane Π containing point P is considered. The
C

intersection of the hydrostatic stress axis with said plane defines point A.
©

Definition 8.3. Based on Figure 8.5, the normal octahedral stress is


defined as   √
OA = 3 σoct
and the shear or tangential octahedral stress is
  √
AP = 3 τoct .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Principal Stress Space 377

rs
ee
s gin
t d le En
Figure 8.5: Definitions of the normal and shear octahedral stresses.

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
Remark 8.4. The normal octahedral stress σoct informs of the dis-

a
i
an an n

tance between the origin O of the principal stress space and the oc-
y ha

tahedral plane that contains point P. The locus of points in the prin-
√ the same value of σoct is the octahedral plane
le
cipal stress space with
liv or ec

placed at a distance 3 σoct of the origin.


M

.A

The shear octahedral stress τoct informs of the distance between


point P and the hydrostatic stress axis. It is, thus, a measure of the
m

distance that separates the stress state characterized by point P from


uu
e

a hydrostatic stress state. The locus of points in the principal stress


X Th

space with the same value of τoct is a cylinder


er

whose axis is the hy-


tin


drostatic stress axis and whose radius is 3 τoct .
on

.O
C

 
©

The distance OA can be computed as the projection of the vector OP on the
unit normal vector of the octahedral plane, n,
⎡ √ ⎤ ⎫
1/ 3 √ ⎪

  √ ⎪
not  ⎢ √ ⎥ ⎪
(σ1 + σ2 + σ3 ) = 3σm ⎪
OA = OP · n ≡ σ1 , σ2 , σ3 ⎣ 1/ 3 ⎦ = 3

√ 3 ⇒
1/ 3 ⎪

  √ ⎪

OA = 3 σoct ⎪

(8.15)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
378 C HAPTER 8. P LASTICITY

I1
σoct = σm = (8.16)
3

 (8.9) of mean stress σm has been taken into account.


where the definition
The distance AP can be obtained solving for the right triangle OAP in Fig-
ure 8.5,
 2
AP = OP2 − OA2 = σ12 + σ22 + σ32 − 1 (σ1 + σ2 + σ3 )2 . (8.17)
3
By means of several algebraic operations, this distance can be expressed in terms

rs
of the second invariant of the deviatoric stress tensor in (8.14), J2 , as

ee
 2   √ ⎫
AP = 2 J  =⇒ AP = 2 (J  )1/2 ⎬

s gin
2 2
  √ =⇒ (8.18)
AP = 3 τ ⎭

t d le En
oct

r


ba
ge ro or
eS m
2  1/2
τoct = (J )

ci
(8.19)
f 3 2

ra
C d P cs
b
a
i
Alternative expressions of τoct in terms of the value of J2 in (8.14) are
an an n
y ha

 1/2
1 1
le
2
τoct = √ σ1 + σ2 + σ3 − (σ1 + σ2 + σ3 )
2 2 2
liv or ec

and
3 3
(8.20)
M

.A

1  1/2
τoct = √ (σ1 − σ2 )2 + (σ2 − σ3 )2 + (σ1 − σ3 )2 .
m

3 3
d
uu
e
X Th

er
tin

Remark 8.5. In a pure spherical stress state of σ ,


on

.O

σ = σ sph = σm 1 ⇔ σ  = σ −σ
σ sph = 0 ⇔ J2 = 0 ⇔ τoct = 0 .
C

A spherical stress state is characterized by τoct = 0 and, thus, is lo-


cated on the hydrostatic stress axis (see Figure 8.5).
In a pure deviatoric stress state of σ ,

σ = σ  ⇔ σm = Tr (σ σ ) = Tr σ  = 0 ⇔ σoct = 0 .

A deviatoric stress state is characterized by σoct = 0 and, therefore, is


located on the octahedral plane containing the origin of the principal
stress space.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Principal Stress Space 379

Remark 8.6. A point P of the principal stress space is univocally


characterized by the three invariants I1 ≡ J1 , J2 and J3 (see Fig-
ure 8.6):

− The first stress invariant I1 characterizes the distance (= 3 σoct )
from the origin to the octahedral plane Π containing this point
through the relation σoct = I1 /3. Thus, it places point P in a cer-
tain octahedral plane.
− The second 
√ deviatoric stress invariant J2 characterizes the dis-

rs
tance (= 3 τoct ) from the hydrostatic stress axis to the point.

ee
Thus, it places point P on a certain circle in the octahedral

s gin
plane with center in the hydrostatic stress axis and radius
√ √
3 τoct = 2 (J2 )1/2 .

t d le En
− The third deviatoric stress invariant J3 characterizes the position

r
ba
of the point on this circle by means of an angle θ (J3 ).

ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu

hydrostatic stress axis


e
X Th

er
tin
on

.O
C

Figure 8.6: Univocal definition of a point by means of the invariants I1 , J2 and J3 .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
380 C HAPTER 8. P LASTICITY

Remark 8.7. Figure 8.7 shows the projection of the principal stress
space on an octahedral plane Π . The division of the stress space
into six sectors can be observed in this projection. Each sector is
characterized by a different ordering of the principal stresses and the
sectors are separated by the projections on the plane of the bisectors
σ2 = σ3 , σ1 = σ3 and σ1 = σ2 .
Selecting the criterion σ1 ≥ σ2 ≥ σ3 automatically reduces the fea-
sible work domain to the sector marked in gray in the figure. The
intersection of any surface of the type f (σ1 , σ2 , σ3 ) = 0 with the

rs
plane Π is reduced to a curve in said sector.

ee
This curve can be automatically extended to the rest of sectors, that

s gin
is, the curve obtained with the same function f (σ1 , σ2 , σ3 ) = 0 but
considering the different orderings of the principal stresses can be

t d le En
easily plotted, by considering the symmetry conditions with respect
to the bisector planes. The resulting curve presnts, thus, three axes

r
ba
ge ro or
of symmetry with respect to each of the axis in Figure 8.7.

eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 8.7: Projection on an octahedral plane.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Rheological Models 381

8.4 Rheological Models


Rheological models are idealizations of mechanical models, constructed as a
combination of simple elements, whose behavior is easily intuitable, and that al-
low perceiving more complex mechanical behaviors. Here, as a step previous to
the analysis of elastoplastic models, frictional rheological models will be used to
introduce the concept of irrecoverable or permanent strain and its consequences.

8.4.1 Elastic Element (Spring Element)


The elastic rheological model is defined by a spring with constant E (see Fig-

rs
ure 8.8). The model establishes a proportionality between stress and strain, both

ee
in loading and unloading, being the constant E the proportionality factor (see
Figure 8.8).

s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

Figure 8.8: Stress-strain relation in an elastic model.


d
uu
e
X Th

er
tin

8.4.2 Frictional Element


on

.O

Consider a solid block placed on a rough surface (see Figure 8.9) and subjected
C

to a vertical compressive load N and a horizontal load F (positive rightward and


©

negative leftward). δ is the horizontal displacement of the block. The Coulomb


friction model2 establishes that the modulus of the reaction force R exerted by
the contact surface on the block cannot exceed a certain limit value Fu = μ N,
where μ ≥ 0 is the friction coefficient between the block and the surface. Con-
sequently, while the load F is below said limit value, the block does not move.
When the limit value Fu = μ N is reached, the block starts moving in a quasi-
static state (without any acceleration). To maintain the quasi-static regime, this
limit value must no be exceeded. These concepts can be mathematically ex-
pressed as

2 The Coulomb friction model is also known as dry friction model.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
382 C HAPTER 8. P LASTICITY

|F| < μ N ⇐⇒ δ =0 there is no motion ,


|F| = μ N ⇐⇒ δ
= 0 there is motion , (8.21)
|F| > μ N impossible .

The behavior of the Coulomb friction model, in terms of the force-displacement


relation F − δ , is graphically represented in Figure 8.9, both for positive values
of the load F (rightward motion) and negative ones (leftward motion).
By analogy with the mechanical friction model, the frictional rheological
model in Figure 8.10 is defined, where σ is the stress (analogous to the load
F in the Coulomb model) that acts on the device and ε is the strain suffered by

rs
this device (analogous to the displacement δ ). This rheological model includes

ee
a frictional device characterized by a limit value σe (analogous to the role of μ N

s gin
in the Coulomb model) whose value cannot be exceeded.
Figure 8.11 shows the stress-strain curve corresponding to the frictional rhe-

t d le En
ological model for a loading-unloading-reloading cycle, which can be split into
the following sections.

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e

Figure 8.9: Coulomb’s law of friction.


X Th

er
tin
on

|σ | < σe → Δ ε = 0
.O

|σ | = σe → Δ ε
= 0
C

|σ | > σe → impossible
©

Figure 8.10: Frictional rheological model.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Rheological Models 383

Section 0 − 1: The (tensile) stress σ in-


creases until the threshold value σ = σe
is reached. There is no strain.
Section 1 − 2: Once the threshold
σ = σe has been reached, stress can-
not continue increasing although it can
keep its value constant. Then, the fric-
tional element flows, generating a strain
ε that grows indefinitely while the stress
is maintained (loading process).

rs
Section 2 − 3: At point 2, the tendency

ee
of the stress is inverted, stress starts de-

s gin
creasing (Δ σ < 0) and unloading be- Figure 8.11: Stress-strain curve
gins (σ < σe ). Further strain increase is for a loading-unloading-reloading

t d le En
automatically halted (Δ ε = 0). This sit- cycle in a frictional rheological
uation is maintained until stress is can- model.

r
ba
celed (σ = 0) at point 3. Note that, if

ge ro or
eS m
ci
the process was to be halted at this point, the initial state of null stress would be
f

ra
recovered but not the initial state of null strain. Instead, a permanent or residual
C d P cs
b
a
strain would be observed (ε
= 0). This reveals that, in this model, the trajec-
i
an an n

tory of the stress-strain curve is different in the loading and unloading regimes
y ha

and that the deformation process is (from a thermodynamic point of view) irre-
le
versible in character.
liv or ec
M

.A

Section 3 − 4: Beyond point 3, the sign of the stress is inverted and stress be-
comes compressive. However, since |σ | < σe , no changes in strain are observed
m

(Δ ε = 0).
d
uu
e

Section 4−5: At point 4, the criterion |σ | = σe is satisfied and the model enters a
X Th

er
tin

loading regime again. The element flows at a constant stress value σ = −σe , gen-
erating negative strain (Δ ε < 0), which progressively reduces the accumulated
on

.O

strain. Finally, at point 5, the initial strain state is recovered, but not the original
stress state. Beyond this point, if unloading was imposed, there would be a cor-
C

responding decrease in stress until the cycle was closed at point 0. Conversely,
the loading regime could continue, generating a permanent negative strain.

8.4.3 Elastic-Frictional Model


The basic rheological elements, elastic and frictional, can be combined to pro-
duce a more complex model, named elastic-frictional model, by placing an elas-
tic element, characterized by the parameter E, in series with a frictional ele-
ment, characterized by the parameter σe (denoted as elastic limit), as shown in
Figure 8.12. Consider σ is the stress that acts on the model and ε is the to-
tal strain of this model. Since the basic elements are placed in series, the same

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
384 C HAPTER 8. P LASTICITY

stress will act on both of them. On the other hand, the total strain can be decom-
posed into the sum of the strain experienced by the elastic element (ε e ) plus the
strain experienced by the frictional device ε f . The same logic can be applied
at incremental level.

σ = σe = σ f ⎫
σ
ε = εe + ε f = + ε f ⎬ Additive decomposition (8.22)
E
Δ ε = Δ εe + Δ ε f ⎭ of strain

rs
ee
s gin
Frictional element
Elastic element

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec

Figure 8.12: Elastic-frictional element.


M

.A
m

Taking into account the stress-strain behavior of each basic element that com-
d
uu

poses the rheological model, the combined model will satisfy:


e
X Th


er
tin

Δ ε = Δ εe
• |σ | < σe =⇒ Δ ε = 0 =⇒ Δ ε = Δ ε =⇒
f e
on

.O

Δ σ = EΔ ε
C

The frictional element does not deform for stresses |σ | < σe , therefore all
©

strains are absorbed by the elastic element.

σ
• |σ | = σe =⇒ Δ ε f
= 0 =⇒ ε = + ε f =⇒
E

|σ | = σe
Δ ε = Δ ε f =⇒ Δ ε e = 0 =⇒ Δ σ = 0

All strain increments are absorbed by the frictional element with a null in-
crement of stress.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Rheological Models 385

• |σ | > σe

This is incompatible with the characteristics of the frictional element.

Figure 8.13 shows the stress-strain curve for a loading-unloading-reloading


cycle of the elastic-frictional model, which can be decomposed into the follow-
ing sections.
Section 0 − 1:
|σ | < σe =⇒ Δ ε f = 0 =⇒ Δ ε = Δ ε e

rs
This section corresponds to the elastic loading phase. At the end of the loading,
at point 1, the strain is ε = ε e = σe /E. The value of σe at the end of this elastic

ee
section justifies its denomination as elastic limit.

s gin
Section 1 − 2:

t d le En
 σe
ε= +ε f

r
|σ | = σe =⇒ Δ ε
= 0 =⇒
f

ba
E

ge ro or
eS m
Δε = Δε f > 0

ci
f

ra
C d P cs
b
This section corresponds to the frictional loading during which no deformation

a
i
is generated in the elastic element (no elastic strain is generated) and all incre-
an an n
y ha

ments of strain are absorbed by the frictional element.


le
Section 2 − 3:
liv or ec

|σ | < σe =⇒ Δ ε f = 0 =⇒ Δ ε = Δ ε e
M

.A

This section corresponds to the elastic unloading. At the end of the unloading,
m

at point 3, the initial state of null stress is recovered (σ = 0). Consequently, the
d

elastic strain at this point is ε e = σ /E = 0 and, thus, the residual or irrecoverable


uu
e

strain is ε = ε f
= 0. That is, the strain generated by the frictional element during
X Th

er
tin

the frictional loading section 1 − 2 is not recovered during this phase of stress
relaxation to zero. This allows qualifying the frictional component of strain ε f
on

.O

as an irrecoverable or irreversible strain.


C

Section 3 − 4:
©

|σ | < σe =⇒ Δ ε f = 0 =⇒ Δ ε = Δ ε e
This section corresponds to the elastic reloading phase, similar to section 0 − 1
but with a compressive stress (σ < 0). The frictional component of strain is not
modified during the reloading and the final value, at point 4, of the elastic strain
is ε e = −σe /E.
Section 4 − 5:

⎨ ε = − σe + ε f
|σ | = σe =⇒ Δ ε =
f

0 =⇒ E
⎩Δε = Δε f < 0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
386 C HAPTER 8. P LASTICITY

This section corresponds to the


frictional reloading during which
negative frictional strain is gen-
erated Δε f < 0 . Therefore,
the total value of the frictional
strain decreases until it becomes
zero at point 5 (characterized by
ε = ε e = −σe /E and ε f = 0). An
additional elastic unloading at this
point would result in recovering the

rs
initial state 0.

ee
Figure 8.13: Stress-strain curve for
a loading-unloading-reloading cycle in

s gin
an elastic-frictional rheological model.

t d le En

r
8.4.4 Frictional Model with Hardening

ba
ge ro or
eS m
ci
Consider the rheological model in Figure 8.14 composed of an elastic element
f

ra
(characterized by the parameter H  , which will be denoted as hardening mod-
C d P cs
b
a
ulus) and a frictional element (characterized by the elastic limit σe ) placed in
i
an an n

parallel. The parallel arrangement results in both rheological elements sharing


y ha

the same strain, while the total stress in the model is the sum of the stress in the
frictional element (σ (1) ) plus the stress in the elastic element (σ (2) ).
le
liv or ec
M

.A


σ = σ (1) + σ (2)
m

Δ σ = Δ σ (1) + Δ σ (2) (8.23)


uu
e

ε= εe = εf
X Th

er
tin
on

.O
C

Figure 8.14: Frictional model with hardening.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Rheological Models 387

Analyzing separately the behavior of each element results in:


a) Frictional element
 
 (1) 
σ  < σe Δε f = Δε = 0
 
 (1) 
 σ  = σe Δ ε f = Δ ε
= 0 (8.24)
 
 (1) 
σ  > σe impossible

rs
ee
b) Elastic element

s gin

σ (2) = H  ε e = H  ε

t d le En
(8.25)
Δ σ (2) = H  Δ ε e = H  Δ ε

r
ba
ge ro or
eS m
c) Combining (8.24) and (8.25) leads to

ci
     f

ra

C d P cs
 (1)   (2) 
 = σ − H  ε 
b
a
σ =
  σ − σ (8.26)
i
an an n
y ha

In agreement with (8.24) and (8.25), the following situations can be estab-
le
lished regarding the rheological model:
liv or ec


 
M

.A

 (1)    Δε f = Δε = 0
• σ  < σe ⇐⇒ σ − H ε  < σe =⇒

Δ σ (2) = H  Δ ε e = H  Δ ε = 0
m


uu

Δ σ = Δ σ (1)
e
X Th

=⇒
er
tin

Δε = 0
on

.O

All the stress is absorbed by the frictional device and strain is null.
C

⎧ 

⎨ σ (1)  = σe
©

   
 
• σ (1)  = σe ⇐⇒ σ − H  ε  = σe =⇒    
⎩ σ (2)  = σ − σ (1) 

=⇒ Δ σ (2) = Δ σ = H  Δ ε
All stress increments are totally absorbed by the elastic element.

Figure 8.15 shows the stress-strain curve corresponding to this rheological


model for a loading-unloading-reloading cycle, which can be decomposed into
the following sections.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
388 C HAPTER 8. P LASTICITY

Section 0 − 1:

  Δ σ (2) = EΔ ε = 0
 (1) 
σ  < σe =⇒ Δ ε = 0 =⇒
Δ σ (1) = Δ σ

In this section all the stress is absorbed by the frictional element. At the end of
the section, at point 1, the strain is ε = 0 and the stress is σ = σe . This section
is characterized by the condition
 
σ − H  ε  < σe .

rs
ee
Section 1 − 2:

s gin

  σ = σe + σ (2)
 (1) 
σ  = σe =⇒

t d le En
Δ σ = Δ σ (2) = H  Δ ε

r
ba
ge ro or
eS m
This is a loading section in which all stress is absorbed by the elastic element.

ci
f
In global terms, the model increases its capacity to resist stress (the model is

ra
C d P cs
said to suffer hardening) proportionally to the increment of strain, being the
b
a
proportionality factor the hardening modulus H  . This section is characterized
i
an an n

by the condition
y ha

 
σ − H  ε  = σe .
le
liv or ec

Section 2 − 3:
M

.A


  Δ σ (1) = Δ σ
m

 (1) 
d

σ  < σe =⇒ Δ ε = 0 =⇒
uu

Δ σ (2) = 0
e
X Th

er
tin

In this section the stress in the frictional element decreases with a null incre-
on

.O

ment of strain and keeping the stress constant in the elastic element. This state
is maintained until stress is totally inverted in the frictional element. Thus, at
C

point 3, the stress is σ (1) = −σ e . This section is characterized by the condition


©

 
σ − H  ε  < σe .

Section 3 − 4:

  σ = −σe + σ (2)
 (1) 
 σ  = σe =⇒
 Δ σ = Δ σ (2) = H  Δ ε
−σ e
The situation is symmetrical with respect to section 1 − 2, with the elastic ele-
ment decreasing the stress it can bear, until the stress becomes null at point 3,

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Rheological Models 389

where σ (1) = −σe and σ (2) = 0. This section is characterized by the condition
 
σ − H  ε  = σe .

Beyond this point, relaxation of the stress in the frictional element leads to the
original state at point 0.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec

Figure 8.15: Stress-strain curve for a loading-unloading-reloading cycle in a frictional


M

.A

rheological model with hardening.


m

d
uu
e
X Th

8.4.5 Elastic-Frictional Model with Hardening


er
tin

Combining now an elastic element, with elastic modulus E, in series with the
on

.O

frictional model introduced in section 8.4.4, which has a hardening modulus H 


and an elastic limit σe , the elastic-frictional model with hardening shown in
C

Figure 8.16 is obtained.


©

Applying the stress equilibrium and strain compatibility equations on the


model (see Figure 8.16) results in


ε = εe + ε f
→ Additive decomposition
Δε = Δ εe + Δ ε f of strain
 (8.27)
σ = σe = σ f
Δσ = Δσe = Δσ f

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
390 C HAPTER 8. P LASTICITY

rs
ee
Figure 8.16: Elastic-frictional model with hardening.

s gin
t d le En
where σ e and σ f represent, respectively, the stresses sustained by the elastic

r
ba
element and the frictional model with hardening. Combining now the behav-

ge ro or
eS m
ci
ior of an elastic element (see Figure 8.8) with that of the frictional model with
f

ra
hardening in Figure 8.14, yields the following situations:
C d P cs
b
a

i
  Δε f = 0
an an n

• σ − H ε f  < σe =⇒

=⇒ Δ σ = EΔ ε
y ha

Δ σ = Δ εe
le
liv or ec

The frictional element with hardening does not deform and the increment of
M

.A

strain Δ ε is completely absorbed by the elastic element. This case is denoted


m

as elastic process.
d

 
uu

• σ − H  ε f  = σe
e
X Th

er
tin

⎧ 
⎨ σ > 0 and Δ σ > 0
on

Δ σ = Δ σ f = H Δ ε f
.O

a) σ Δ σ > 0 ⇐⇒ or =⇒
⎩ Δ σ = Δ σ e = EΔ ε e
C

σ < 0 and Δ σ < 0


©

1 1 E + H
=⇒ Δ ε = Δ ε e + Δ ε f = Δσ +  Δσ = Δσ
E H EH 

⎨ Δ σ = Ee f Δ ε
=⇒ 
⎩ Ee f = E H
E + H
The strain increment is absorbed by the two elements of the model (the
frictional one with hardening and the elastic one). The relation between
the stress increment Δ σ and the strain increment Δ ε is given by the

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Rheological Models 391

elastic-frictional tangent modulus E e f . This case is called inelastic load-


ing process.

⎨ σ > 0 and Δ σ < 0
b) σ Δ σ < 0 ⇐⇒ or

σ < 0 and Δ σ > 0
=⇒ Δ ε f = 0 =⇒ Δ ε = Δ ε e =⇒ Δ σ = E Δ ε
Every strain increment Δ ε is absorbed by the elastic element. This case
is named elastic unloading process.

rs
ee
Figure 8.17 shows the stress-strain curve corresponding to the model for a
loading-unloading-reloading cycle, in which the following sections can be dif-

s gin
ferentiated.

t d le En
Section 0 − 1 and section 2 − 3:
 

r
σ − H  ε f  < σe =⇒ Δ σ = EΔ ε

ba
ge ro or
eS m
ci
f

ra
Correspond to elastic processes.
C d P cs
b
a
Section 1 − 2 and section 3 − 4:
i
an an n

 
y ha

 σ − H  ε f  = σe
le
=⇒ Δ σ = E e f Δ ε
liv or ec

σ Δσ > 0
M

.A

Correspond to inelastic loading processes.


m


uu

Point 2: 
e

σ − H  ε f  = σe
X Th

er

=⇒ Δ σ = EΔ ε
tin

σ Δσ < 0
on

.O

Corresponds to an elastic unloading process.


C

Note that if H  = 0, then E e f = 0, and the elastic-frictional model in Fig-


©

ure 8.13 is recovered.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
392 C HAPTER 8. P LASTICITY

Elastic loading

Elastic

rs
ee
s gin
Figure 8.17: Stress-strain hardening curve for a loading-unloading-reloading cycle in an
elastic-frictional model with hardening.

t d le En

r
ba
8.5 Elastoplastic Phenomenological Behavior

ge ro or
eS m
ci
f
Consider a steel bar of length  and cross-section A subjected to a tensile force F

ra
C d P cs
at its extremes. The stress in the bar will be σ = F/A (see Figure 8.18) and the
b
a
i
corresponding strain can be estimated as ε = δ /, where δ is the lengthening
an an n
y ha

of the bar. If the bar is subjected to several loading and unloading cycles, the
response typically obtained, in terms of stress-strain curve σ − ε, is as indicated
le
liv or ec

in Figure 8.19.
M

.A

Observation of the first cycle reveals that, as long as the stress does not ex-
ceed the value σe (denoted as elastic limit) in point 1, the behavior is linear
m

elastic, characterized by the elastic modulus E (σ = Eε), and there do not ex-
d
uu

ist irrecoverable strains (in a possible posterior unloading, the strain produced
e
X Th

during loading would be recovered).


er
tin

For stress values above σe , the behavior ceases to be elastic and part of the
strain is no longer recovered during an ensuing unloading to null stress (point 3).
on

.O

There appears, thus, a remaining strain named plastic strain, ε p . However, dur-
C

ing the unloading section 2 − 3 the behavior is again, in an approximate form,


©

incrementally elastic (Δ σ = E Δ ε). The same occurs with the posterior reload-
ing 3 − 2, which produces an incrementally elastic behavior, until the stress
reaches, in point 2, the maximum value it will have achieved during the loading

Figure 8.18: Uniaxial tensile loading test.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Elastoplastic Phenomenological Behavior 393

second unloading
first unloading

rs
ee
s gin
t d le En
Figure 8.19: Response to loading-unloading-reloading cycles in an uniaxial tensile load-
ing test.

r
ba
ge ro or
eS m
ci
f

ra
process. From this point on, the behavior is no longer incrementally elastic (as
C d P cs
b
a
if the material remembered the maximum stress to which it has been previously
i
an an n

subjected). A posterior loading-unloading-reloading cycle 2 − 4 − 5 − 4 exposes


y ha

again that, during section 2 − 4, additional plastic strain is generated, which ap-
le
pears in the form of permanent strain in point 5, and, also, additional elastic
liv or ec

strain ε e is produced, understood as the part of the strain that can be recovered
M

.A

during the unloading section 4 − 5.


m

8.5.1 Bauschinger Effect


uu
e
X Th

Consider a sample of virgin material (a material that has not suffered previous
er
tin

states of inelastic strain) subjected to an uniaxial tensile test and another sample
on

of the same virgin material subjected to an uniaxial compressive test. In certain


.O

materials, the responses obtained, in terms of the stress-strain curve σ − ε in


C

Figure 8.20, for both tests are symmetrical with respect to the origin. That is, in
©

the tensile test the response is elastic up to a value of σ = σe (tensile elastic limit)
and in the compressive response the answer is also elastic up to a value of σ =
−σe (compressive elastic limit), being the rest of both curves (for an assumed
regime of monotonous loading) also symmetrical. In this case, the stress-strain
curve of the virgin material is said to be symmetrical in tension and compression.
Suppose now that a specimen that has been previously subjected to a his-
tory of plastic strains3 , for example a tensile loading-unloading cycle such as
the 0 − 1 − 2 − 3 cycle shown in Figure 8.19, undergoes now a compressive
test. Consider also that σy > σe is the maximum stress the material has been
3 This procedure is known as cold hardening and its purpose is to obtain an apparent elastic
limit that is superior to that of the virgin material σy > σe .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
394 C HAPTER 8. P LASTICITY

subjected to during the loading process. An hypothetical symmetrical behavior


would result in the material having now an elastic behavior in the stress range
[−σy , σy ]. However, in certain cases, the elastic behavior in compression ends
much earlier (see Figure 8.20). This is the effect known as Bauschinger effect
or kinematic hardening. Note that the stress-strain curve of the elastic-frictional
model in Figure 8.17 exhibits this type of hardening.

curve of the virgin material

rs
ee
s gin
curve of the stretched material

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
curve without the Bauschinger effect
b
a
i
an an n
y ha

le
liv or ec
M

.A

Figure 8.20: Bauschinger effect or kinematic hardening.


m

d
uu
e
X Th

er
tin

Remark 8.8. In view of the phenomenological behavior observed in


on

.O

Figure 8.19 and in Figure 8.20, the elastoplastic behavior is charac-


terized by the following facts:
C

1) Unlike in the elastic case, there does not exist unicity in the
stress-strain relation. A same value of strain can correspond to
infinite values of stress and vice-versa. The stress value depends
not only on the strain, but also on the loading history.
2) There does not exist a linear relation between stress and strain.
At most, this linearity may be incremental in certain sections of
the deformation process.
3) Irrecoverable or irreversible strains are produced in a loading-
unloading cycle.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Incremental Theory of Plasticity in 1 Dimension 395

8.6 Incremental Theory of Plasticity in 1 Dimension


The elastoplastic behavior analyzed in sec-
tion 8.5 can be modeled using mathemati- elastoplastic region
cal models of certain complexity4 . One of
the most popular approximations is the so-
called incremental theory of plasticity. In a
one-dimensional case, the theory seeks to
approximate a stress-strain behavior such
as the one observed in Figure 8.19 by

rs
means of piece-wise approximations using
elastic region
elastic and inelastic regions such as the

ee
ones shown in Figure 8.21. The generaliza-

s gin
tion to several dimensions requires the in- Figure 8.21: Uniaxial stress-strain
troduction of more abstract concepts.

t d le En
curve for an elastoplastic model.

r
ba
ge ro or
eS m
8.6.1 Additive Decomposition of Strain. Hardening Variable

ci
f

ra
The total strain ε is decomposed into the sum of an elastic (or recoverable)
C d P cs
b
a
strain ε e , governed by Hooke’s law, and a plastic (or irrecoverable) strain ε p ,
i
an an n
y ha

⎧ ⎧

⎨ ⎪

le
Additive decomposition ε = ε + ε =⇒ dε = dε + dε
e p e p
liv or ec

σ dσ (8.28)
⎪ ⎪
M

.A

of strain ⎩ εe = ⎩ dε e =
E E
m

d
uu
e

where E is the elastic modulus. In addition, the hardening variable α (σ , ε p ) is


X Th

er
tin

defined by means of the evolution equation as follows5 .


on

.O


⎨ dα = sign (σ ) dε
⎪ p
C

Hardening dα ≥ 0 (8.29)
©

variable α ⎪
⎩ 
α p =0
ε =0

4 Up to a certain point, these models may be inspired, albeit with certain limitations, in the
elastic-frictional rheological models described in section 8.4.
5 Here, the sign operator is used, which is defined as x ≥ 0 ⇐⇒ sign (x) = +1 and
x < 0 ⇐⇒ sign (x) = −1.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
396 C HAPTER 8. P LASTICITY

Remark 8.9. Note that the hardening variable α is always positive, in


agreement with its definition in (8.29), and, considering the modules
of the expression dα = sign (σ ) dε p , results in

dα = |dα| = |sign (σ )| |dε p | =⇒ dα = |dε p | .


  
=1
Then, for a process with monotonously increasing plastic strains,

rs
both variables coincide,

ee
εp εp

s gin
dε p ≥ 0 =⇒ α = |dε p | = dε p = ε p .

t d le En
0 0

However, if the process does not involve a monotonous increase, the

r
ba
ge ro or
plastic strain may decrease and its value no longer coincides with

eS m
ci
that of the hardening variable α.
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec

8.6.2 Elastic Domain. Yield Function. Yield Surface


M

.A

The elastic domain in the stress space is defined as the interior of the domain
enclosed by the surface F (σ , α) = 0,
m

d
uu
e

Elastic domain: Eσ := {σ ∈ R | F (σ , α) < 0} (8.30)


X Th

er
tin
on

.O

where the function F (σ , α) : R × R+ → R is denoted as yield function.


C

The initial elastic domain E0σ is defined as the elastic domain corresponding
©

to a null plastic strain (ε p = α = 0),

Initial elastic domain: E0σ := {σ ∈ R | F (σ , 0) < 0} . (8.31)

An additional requirement of the initial elastic domain is that it must contain the
null stress state,
0 ∈ E0σ =⇒ F (0, 0) < 0 , (8.32)
and this is achieved by defining a yield function of the type

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Incremental Theory of Plasticity in 1 Dimension 397

Yield function: F (σ , α) ≡ |σ | − σy (α) (8.33)

where σy (α) > 0 is known as the yield stress. The initial value (for α = 0)
of the yield stress is the elastic limit σe (see Figure 8.22) and the function
σy (α) : R+ → R+ is named hardening law.

rs
ee
hardening

s gin
parameter

t d le En

r
ba
ge ro or
eS m
admissible

ci
f

ra
stress space
C d P cs
b
a
i
an an n
y ha

le
liv or ec

Figure 8.22: Hardening law and admissible stress space.


M

.A
m

The yield surface is defined as the boundary of the elastic domain.


d
uu
e

! "
X Th

er
tin

Yield surface: ∂ Eσ := σ ∈ R | F (σ , α) ≡ |σ | − σy (α) = 0 (8.34)


on

.O

The elastic domain Eσ together with its boundary ∂ Eσ determine the admissible
C

stress space (domain) Ēσ


©

Admissible stress space:


# ! " (8.35)
Ēσ = Eσ ∂ Eσ = σ ∈ R | F (σ , α) ≡ |σ | − σy (α) ≤ 0

and it is postulated that any feasible (admissible) stress state must belong to
the admissible stress space Ēσ . Considering the definitions of elastic domain
in (8.30), yield surface in (8.34) and admissible stress space in (8.35), the fol-
lowing is established.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
398 C HAPTER 8. P LASTICITY

$
σ in the elastic domain
F (σ , α) < 0 ⇐⇒ |σ | < σy (α) ⇐⇒
(σ ∈ Eσ )
$
σ on the yield surface (8.36)
F (σ , α) = 0 ⇐⇒ |σ | = σy (α) ⇐⇒
(σ ∈ ∂ Eσ )

F (σ , α) > 0 ⇐⇒ |σ | > σy (α) ⇐⇒ non-admissible stress state

rs
Remark 8.10. Note how, in (8.35), the admissible stress space de-

ee
pends on the hardening variable α. The admissible domain evolves
with the yield function σy (α) such that (see Figure 8.22)

s gin
Ēσ ≡ [−σy (α) , σy (α)] .

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
8.6.3 Constitutive Equation
i
an an n

To characterize the response of the material, the following situations are defined:
y ha

le
• Elastic regime
liv or ec
M

.A

σ ∈ Eσ =⇒ dσ = Edε (8.37)
m

d
uu

• Elastoplastic regime in unloading


e
X Th

%
er
tin

σ ∈ ∂ Eσ
=⇒ dσ = Edε (8.38)
on

.O

dF (σ , α) < 0
C

• Elastoplastic regime in plastic loading


%
σ ∈ ∂ Eσ
=⇒ dσ = E ep dε (8.39)
dF (σ , α) = 0
where E ep is denoted as elastoplastic tangent modulus.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Incremental Theory of Plasticity in 1 Dimension 399

Remark 8.11. The situation σ ∈ ∂ Eσ and dF (σ , α) > 0 cannot oc-


cur since, if σ ∈ ∂ Eσ , from (8.34) results
F (σ , α) ≡ |σ | − σy (α) = 0 .

If, in addition, dF (σ , α) > 0 then,

F (σ + dσ , α + dα) = F (σ , α) + dF (σ , α) > 0
     
=0 >0

rs
ee
and, in agreement with (8.36), the stress state σ + dσ is not admis-

s gin
sible.

t d le En

r
ba
8.6.4 Hardening Law. Hardening Parameter

ge ro or
eS m
ci
The hardening law provides the evolution of the yield stress σy (α) in terms of
f

ra
C d P cs
the hardening variable α (see Figure 8.22). Even though the aforementioned
b
a
i
hardening law may be of a more general nature, it is common (and often suffi-
an an n

cient) to consider a linear hardening law of the type


y ha

le
liv or ec

σy = σe + H  α =⇒ dσy (α) = H  dα , (8.40)


M

.A

where H  is known as the hardening parameter.


m

d
uu

8.6.5 Elastoplastic Tangent Modulus


e
X Th

er
tin

The value of the elastoplastic tangent modulus E ep introduced in (8.39) is calcu-


lated in the following manner. Consider an elastoplastic regime in plastic load-
on

.O

ing. Then, from (8.39)6 ,


C

%
©

σ ∈ ∂ Eσ =⇒ F (σ , α) ≡ |σ | − σy (α) = 0
=⇒
dF (σ , α) = 0 (8.41)
d |σ | − dσy (α) = 0 =⇒ sign (σ ) dσ − H  dα = 0 ,

where (8.40) has been taken into account. Introducing the first expression of
(8.29) in (8.41) yields

1
sign (σ ) dσ − H  sign (σ ) dε p = 0 =⇒ dε p = dσ . (8.42)
H
6 The property d |x|/dx = sign (x) is used here.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
400 C HAPTER 8. P LASTICITY

Consider now the additive decomposition of strain defined in (8.28), which


together with (8.42) results in


dε = dε e + dε p ⎪



⎬  
1 1 1 1 1
dε = dσ
e
=⇒ dε = dσ +  dσ = + dσ =⇒
E ⎪
⎪ E H E H


dε p =  dσ ⎪
1
⎭ (8.43)
H ⎧
1 ⎨ dσ = E ep dε

rs
dσ = dε =⇒  .
1 1 ⎩ E ep = E H

ee
+ E + H
E H

s gin
t d le En
8.6.6 Uniaxial Stress-Strain Curve

r
ba
The constitutive equation defined by expressions (8.37) to (8.39) allows obtain-

ge ro or
eS m
ci
ing the corresponding stress-strain curve for an uniaxial process of loading-
f

ra
unloading-reloading (see Figure 8.23) in which the following sections are ob-
C d P cs
b
a
served.
i
an an n

Section 0 − 1:
y ha

le
|σ | < σe =⇒ σ ∈ Eσ =⇒ Elastic regime
liv or ec
M

.A

From (8.37), dσ = Edε and the behavior is linear elastic, defining an elastic
region in the stress-strain curve.
m

d
uu
e

Section 1 − 2 − 4:
X Th

er

%
tin

F (σ , α) ≡ |σ | − σy (α) = 0 =⇒ σ ∈ ∂ Eσ Elastoplastic regime


=⇒
on

.O

dF (σ , α) = 0 in plastic loading
C

From (8.39), dσ = E ep dε, defining an elastoplastic region.


©

Section 2 − 3 − 2:

F (σ , α) ≡ |σ | − σy (α) < 0 =⇒ σ ∈ ∂ Eσ =⇒ Elastic regime


From (8.37), dσ = Edε and the behavior is linear elastic, defining an elastic
region in the stress-strain curve.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Incremental Theory of Plasticity in 1 Dimension 401

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
Figure 8.23: Uniaxial stress-strain curve for a loading-unloading-reloading cycle con-
b
a
sidering the incremental theory of plasticity.
i
an an n
y ha

le
liv or ec
M

.A

Remark 8.12. In point 2 of Figure 8.23 the following two processes


are distinguished:
m

%
d

F (σ , α) ≡ |σ | − σy (α) = 0 =⇒ σ ∈ ∂ Eσ
uu

Elastic unloading
e

in section 2 − 3
X Th

dF (σ , α) < 0
er
tin

%
F (σ , α) ≡ |σ | − σy (α) = 0 =⇒ σ ∈ ∂ Eσ
on

.O

Plastic loading in
dF (σ , α) = 0 section 2 − 4
C

Remark 8.13. Note that plastic strain is only generated during the
plastic loading process in the elastoplastic region (see Figure 8.24).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
402 C HAPTER 8. P LASTICITY

Elastoplastic
region

Elastic region

rs
ee
Figure 8.24: Generation of plastic strain in the elastoplastic region.

s gin
t d le En

r
Remark 8.14. Note the similarity between the stress-strain curve in

ba
ge ro or
eS m
Figure 8.23 and the one obtained with the elastic-frictional rheologi-

ci
f
cal model with hardening in section 8.4.5 (Figure 8.17). The friction

ra
C d P cs
strain in said model is equivalent to the plastic strain in the incre-
b
a
i
mental theory of plasticity.
an an n
y ha

le
liv or ec
M

.A

Remark 8.15. The hardening parameter H  plays a fundamental role


m

in the definition of the slope E ep of the elastoplastic region. Follow-


d
uu
e

ing (8.43),
X Th

H
er
tin

E ep = E
E + H
on

.O

and, depending on the value of H  , different situations arise (see Fig-


C

ure 8.25):
©

− H  > 0 =⇒ E ep > 0 → Plasticity with strain hardening. The limit


case H  = ∞ ⇒ E ep = E recovers the linear elastic behavior.
− H  = 0 =⇒ E ep = 0 → Perfect plasticity.
− H  < 0 =⇒ E ep < 0 → Plasticity with strain softening7 . The
limit case corresponds to H  = −E ⇒ E ep = −∞.

7 Plasticity with strain softening presents a specific problematic regarding the uniqueness of
the solution to the elastoplastic problem, which is beyond the scope of this text.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Plasticity in 3 Dimensions 403

Figure 8.25: Role of the hardening parameter H  in the definition of the slope E ep .

rs
ee
8.7 Plasticity in 3 Dimensions

s gin
The incremental theory of plasticity developed in one dimension in section 8.6

t d le En
can be generalized to a multiaxial stress state (three dimensions) using the same

r
ingredients, that is:

ba
ge ro or
eS m
ci
1) Additive decomposition of strain
f

ra
C d P cs
 
b
a
Additive ε = εe +ε p dεε = dεε e + dεε p
i
an an n

decomposition =⇒ (8.44)
y ha

of strain ε e = C −1 : σ dεε e = C −1 : dσσ


le
liv or ec
M

.A

where C −1 is now the (constant) constitutive elastic tensor defined in chapter 6.


m

2) Hardening variable α and flow rule (evolution equations)


d
uu
e


X Th

er

σ , α)
⎨ dε p = λ ∂ G (σ
tin


Flow rule ∂σ (8.45)
on

.O


⎩ dα = λ α ∈ [0, ∞)
C

σ , α) is the plastic potential function.


where λ is the plastic multiplier and G (σ

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
404 C HAPTER 8. P LASTICITY

3) Yield function. Elastic domain and yield surface



σ , α) ≡ φ (σ
F (σ σ ) − σy (α)
Yield function
σy (α) = σe + H  α (hardening law)

σ , α) < 0}
Elastic domain Eσ := { σ | F (σ

Initial (8.46)
σ , 0) < 0}
E0σ := { σ | F (σ
elastic domain

rs
Yield surface σ , α) = 0}
∂ Eσ := { σ | F (σ

ee
#

s gin
Admissible Ēσ = Eσ σ , α) ≤ 0}
∂ Eσ = { σ | F (σ
stress state

t d le En
where φ (σσ ) ≥ 0 is denoted as the equivalent uniaxial stress, σe is the elastic

r
ba
ge ro or
eS m
limit obtained in an uniaxial test of the material (it is a material property) and

ci
σy (α) is the yield stress. The hardening parameter H  plays the same role as
f

ra
C d P cs
in the uniaxial case and determines the expansion or contraction of the elastic
b
a
domain Eσ , in the stress space, as α grows. Consequently,
i
an an n
y ha

le
liv or ec

H  > 0 =⇒ Expansion → Plasticity with hardening


of Eσ with α
M

.A

Contraction
m

H  < 0 =⇒ → Plasticity with softening (8.47)


d

of Eσ with α
uu
e
X Th

Constant
er
tin

H  = 0 =⇒ elastic
 → Perfect plasticity
domain
Eσ = E0σ
on

.O
C

4) Loading-unloading conditions (Karush-Kuhn-Tucker conditions) and con-


sistency condition

Loading-unloading → λ ≥ 0 ; F (σ
σ , α) ≤ 0 ; λ F (σ
σ , α) = 0
conditions
(8.48)
Consistency σ , α) = 0 ⇒ λ F (σ
→ If F (σ σ , α) = 0
condition

The loading-unloading conditions and the consistency condition are additional


ingredients, with respect to the unidimensional case, which allow obtaining, af-
ter certain algebraic manipulation, the plastic multiplier λ introduced in (8.45).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Yield Surfaces. Failure Criteria 405

8.7.1 Constitutive Equation


Similarly to the uniaxial case, the following situations are differentiated in rela-
tion to the constitutive equation:
• Elastic regime
σ = C : dεε
σ ∈ Eσ =⇒ dσ (8.49)

• Elastoplastic regime in unloading


%

rs
σ ∈ ∂ Eσ
σ = C : dεε
=⇒ dσ (8.50)

ee
σ , α) < 0
dF (σ

s gin
t d le En
• Elastoplastic regime in plastic loading
%

r
ba
σ ∈ ∂ Eσ

ge ro or
eS m
=⇒ dσσ = C ep : dεε (8.51)

ci
σ , α) = 0
dF (σ f

ra
C d P cs
b
a
where C ep is known as the elastoplastic constitutive tensor which, after certain
i
an an n

algebraic operations considering (8.44) to (8.48), is defined as


y ha

le
liv or ec

∂G ∂F
M

.A

⊗ C: :C
σ , α) = C −
C ep (σ ∂σ ∂σ
m

∂F ∂G
d

H + :C :
uu

∂σ ∂σ
e
X Th

(8.52)
er
tin

∂G ∂F
Ci jpq Crskl
∂ σ pq ∂ σrs
on

.O

Cep = Ci jkl − i, j, k, l ∈ {1, 2, 3}


i jkl ∂F ∂G
C

H + C pqrs
∂ σ pq ∂ σrs
©

8.8 Yield Surfaces. Failure Criteria


A fundamental ingredient in the theory of plasticity is the existence of an initial
elastic domain E0σ (see Figure 8.26) which can be written as

σ ) ≡ φ (σ
E0σ := { σ | F (σ σ ) − σe < 0} (8.53)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
406 C HAPTER 8. P LASTICITY

and determines a domain in the stress space delimited by the initial yield sur-
face ∂ E0σ ,
σ ) ≡ φ (σ
∂ E0σ := { σ | F (σ σ ) − σe = 0} . (8.54)
Given that the initial elastic domain contains the origin of the stress space
σ = 0), every loading process in any point of the medium will include an elas-

tic regime (as long as the trajectory of the stresses remains inside E0σ , see Fig-
ure 8.26) that will end at the instant in which said trajectory reaches the yield
surface ∂ E0σ . The initial yield surface plays then the role of indicating the in-
stant of failure (understood as the end of the elastic behavior) independently
of the possible post-failure (plastic) behavior that initiates beyond this instant.

rs
Thus, the importance of the initial yield surface and the interest in formulat-

ee
ing the mathematical equations that adequately determine this surface for the

s gin
different materials of interest in engineering.
With the aim of defining the yield surface independently of the reference

t d le En
system (isotropic material)8 , even if formulated in the principal stress space, its
mathematical equation is typically defined in terms of the stress invariants,

r
ba
ge ro or


eS m
F (σσ ) ≡ F I1 , J2 , J3 ,

ci
(8.55)
f

ra
C d P cs
b
and, since the criterion σ1 ≥ σ2 ≥ σ3 is adopted, its definition only affects the

a
i
an an n

first sector of the principal stress space and can be automatically extended, due
y ha

to symmetry conditions (see Remark 8.7), to the rest of sectors in Figure 8.7.
le
liv or ec
M

.A
m

σ ) − σe = 0}
∂ E0σ := { σ | φ (σ
d
uu
e
X Th

σ ) − σe < 0}
E0σ := { σ | φ (σ
er
tin
on

.O
C

Figure 8.26: Initial elastic domain and initial yield surface.

8 An isotropic elastoplastic behavior is characterized by the fact that the yield surface, under-
stood as an additional ingredient of the constitutive equation, is independent of the reference
system.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Yield Surfaces. Failure Criteria 407

8.8.1 Von Mises Criterion


In the von Mises criterion the yield surface is defined as



Von Mises criterion: σ ) ≡ σ̄ (σ
F (σ σ ) − σe = 3 J2 − σe = 0 (8.56)



σ ) = 3 J2 is the effective stress (see Remark 8.3). An alternative
where σ̄ (σ
expression is obtained taking (8.19) and (8.20) and replacing them in (8.56),

rs
which produces

ee
1  1/2

s gin
σ ) ≡ √ (σ1 − σ2 )2 + (σ2 − σ3 )2 + (σ1 − σ3 )2
F (σ − σe = 0 . (8.57)
2

t d le En
The graphical representation of the von Mises yield surface is shown in Fig-

r
ure 8.27.

ba
ge ro or
eS m
ci
 f

ra
C d P cs
2 
b
a
R= σe 2
i
3 R= σe
an an n

3
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 8.27: Von Mises criterion in the principal stress space.


©

Remark 8.16. Equation (8.56) highlights the dependency of the von


Mises yield surface solely on the second stress invariant J2 . Conse-
quently, all the points of the surface are characterized by the same
value of J2 , which defines a cylinder whose axis is the hydrostatic
stress axis.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
408 C HAPTER 8. P LASTICITY

Remark 8.17. The von Mises criterion is adequate as a failure crite-


rion in metals, in which, typically, hydrostatic stress states (both in
tensile and compressive loading) have an elastic behavior and failure
is due to the presence of deviatoric stress components.

Example 8.2 – Compute the expression of the von Mises criterion for an

rs
uniaxial tensile loading case.

ee
s gin
Solution
An uniaxial tensile loading case is characterized by the stress state

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
⎡ ⎤
b
a
σu 0 0
i
an an n

not ⎢ ⎥
σ ≡⎣ 0 0 0 ⎦ .
y ha

le
0 0 0
liv or ec
M

.A

The effective stress is known to be σ̄ = |σu | (see Example 8.1) and, replacing
in the expression of the von Mises criterion (8.56), yields
m

σ ) ≡ σ̄ (σ
F (σ σ ) − σe = |σu | − σe .
uu
e
X Th

er
tin

Thus, the initial elastic domain is characterized in the same way as in unidi-
mensional plasticity seen in Section 8.6.2, by the condition
on

.O
C

σ ) < 0 =⇒ |σu | < σe


F (σ .
©

Example 8.3 – Compute the expression of the von Mises criterion for a stress
state representative of a beam under composed flexure.

Solution
The stress state for a beam under composed flexure is

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Yield Surfaces. Failure Criteria 409

rs
ee
⎡ ⎤
σx τxy 0

s gin
not ⎢ ⎥ 1
σ ≡ ⎣ τxy 0 0⎦ =⇒ σm = σx =⇒
3

t d le En
0 0 0

r
ba
ge ro or
eS m
⎡ ⎤

ci
2
σx τxy f

ra
0
⎢ 3 ⎥
C d P cs
not ⎢
b ⎥

a
1 ⎢ 1

0 ⎥
i
σ = σ − σx 1 ≡ ⎢ τxy − σx ⎥.
an an n

3 ⎣ 1 ⎦
3
y ha

0 0 − σx
le
liv or ec

3

M

.A

Then, the second stress invariant J2 is computed as


 
m

 1   1 4 2 1 2 1 2 1
J2 = σ : σ = σx + σx + σx + τxy + τxy = σx2 + τxy .
d

2 2 2
uu

2 2 9 9 9 3
e
X Th

er
tin

And the effective stress is obtained for the von Mises criterion,
 
on

.O


σ̄ = 3 J2 = σx2 + 3τxy 2 =⇒ F (σ σ ) < 0 =⇒ σ̄ < σe =⇒
C


σco = σx2 + 3τxy e .
2 <σ


The comparison stress, σco = σxx 2 + 3τ 2 , which can be regarded as a scalar
xy
for comparison with the uniaxial elastic limit σe , is commonly used in the
design standards of metallic structures.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
410 C HAPTER 8. P LASTICITY

8.8.2 Tresca Criterion or Maximum Shear Stress Criterion


The Tresca criterion, also known as the maximum shear stress criterion, states
that the elastic domain ends, for a certain point in the medium, when the maxi-
mum shear stress acting on any of the planes containing this point, τ max , reaches
half the value of the uniaxial elastic limit σe ,
σ1 − σ3 σe
τ max = = . (8.58)
2 2
Figure 8.28 illustrates the failure situation in terms of Mohr’s circle in three
dimensions. In a loading process in which this circle increases starting from

rs
the origin, the elastic behavior ends when the circle with radius τ max becomes

ee
tangent to the straight line τ = τ max = σe /2.

s gin
It follows from (8.58) that the Tresca criterion can be written as

t d le En
σ ) ≡ (σ1 − σ3 ) − σe = 0
F (σ (8.59)

r
Tresca criterion:

ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n

Remark 8.18. It can be verified that the Tresca criterion is written in


y ha

an unequivocal form as a function of J2 and J3 and does not depend


le
on the first stress invariant I1 .
liv or ec

  
M

.A

Tresca criterion: F (σ σ ) ≡ (σ1 − σ3 ) − σe ≡ F J2 , J3


m

d
uu
e
X Th

er
tin
on

.O
C

Figure 8.28: Representation of the Tresca criterion using Mohr’s circle in three dimen-
sions.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Yield Surfaces. Failure Criteria 411

hydrostatic stress axis

von Mises

Tresca

rs
ee
s gin
Figure 8.29: Tresca criterion in the principal stress space.

t d le En

r
ba
ge ro or
Figure 8.29 shows the yield surface corresponding to the Tresca criterion in

eS m
ci
the principal stress space, which results in an hexahedral prism whose axis is the
f

ra
C d P cs
hydrostatic stress axis.
b
a
i
an an n
y ha

Remark 8.19. Since the Tresca criterion does not depend on the first
le
stress invariant (and, therefore, on the stress σoct , see (8.16)), the cor-
liv or ec

responding yield surface does not depend on the distance from the
M

.A

origin to the octahedral plane containing the point (see Remark 8.4).
m

Thus, if a point in the stress space, characterized by its stress invari-


d

ants (I1 , J2 , J3 ), is on said yield surface, all the points in the stress
uu
e

space with the same values of J2 and J3 will also be on this surface.
X Th

er
tin

This circumstance qualifies the yield surface as a prismatic surface


whose axis is the hydrostatic stress axis.
on

.O

On the other hand, the dependency on the two invariants J2 and J3 ,
prevents (unlike in the case of the von Mises criterion) the surface
C

from being cylindrical. In short, the symmetry conditions establish


that the surface of the Tresca criterion be an hexagonal prism in-
scribed in the von Mises cylinder (see Figure 8.29).

Remark 8.20. The Tresca criterion is used to model the behavior of


metals, in a similar manner to the case of the von Mises criterion
(see Remark 8.17).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
412 C HAPTER 8. P LASTICITY

Example 8.4 – Compute the expression of the Tresca criterion for an uniaxial
tensile loading case.

Solution
An uniaxial tensile load case is characterized by the stress state

⎡ ⎤

rs
σu 0 0
not ⎢ ⎥

ee
σ ≡⎣ 0 0 0 ⎦ .

s gin
0 0 0

t d le En
For the case σu ≥ 0 ,
%

r
ba
σ1 = σ u

ge ro or
eS m
=⇒ F (σ1 , σ2 , σ3 ) = (σ1 − σ3 ) − σe = σu − σe = |σu | − σe .

ci
σ3 = 0 f

ra
C d P cs
b
a
For the case σu < 0,
i
an an n

%
y ha

σ1 = 0
le
=⇒ F (σ1 , σ2 , σ3 ) = (σ1 − σ3 ) − σe = −σu − σe = |σu | − σe .
liv or ec

σ 3 = σu
M

.A

And the initial elastic domain is then characterized in the same way as in the
m

one-dimensional plasticity seen in Section 8.6.2, by the condition


uu
e
X Th

er

σ ) < 0 =⇒ |σu | < σe .


tin

F (σ
on

.O
C

8.8.3 Mohr-Coulomb Criterion


The Mohr-Coulomb criterion can be viewed as generalization of the Tresca cri-
terion, in which the maximum shear stress sustained depends on the own stress
state of the point (see Figure 8.30). The yield line, in the space of Mohr’s circle,
is a straight line characterized by the cohesion c and the internal friction angle φ ,
both of which are considered to be material properties,
τ = c − σ tan φ . (8.60)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Yield Surfaces. Failure Criteria 413

rs
ee
s gin
Figure 8.30: Representation of the Mohr-Coulomb criterion using Mohr’s circle in three

t d le En
dimensions.

r
ba
ge ro or
eS m
ci
The end of the elastic behavior (failure) in an increasing load process takes
f

ra
C d P cs
place when the first point in the Mohr’s circle (corresponding to a certain plane)
b
a
reaches the aforementioned yield line.
i
an an n

The shear stress in this plane, τ, becomes smaller as the normal stress σ in the
y ha

plane increases. It therefore becomes obvious that the behavior of the model un-
le
liv or ec

der tensile loading is considerably different to the behavior under compressive


loading. As can be observed in Figure 8.30, the yield line crosses the normal
M

.A

stress axis in the positive side of these stresses, limiting thus the material’s ca-
pacity to withstand tensile loads.
m

To obtain the mathematical expression of the yield surface, consider a stress


uu
e

state for which plasticization initiates. In such case, the corresponding Mohr’s
X Th

er
tin

circle is defined by the major and minor principal stresses and is tangent to the
yield line at point A (see Figure 8.31), verifying
on

.O


⎪ σ − σ3
C


⎨ τA = R cos φ = 1 cos φ
σ1 − σ3
©

R= =⇒ 2

⎩ σA = σ1 + σ3 + R sin φ = σ1 + σ3 + σ1 − σ3 sin φ
2 ⎪
2 2 2
(8.61)
and, replacing (8.61) in (8.60), results in
τA = c − σA tan φ =⇒ τA + σA tan φ − c = 0 =⇒
 
σ1 − σ3 σ1 + σ3 σ1 − σ3
cos φ + + sin φ tan φ − c = 0 =⇒ (8.62)
2 2 2
(σ1 − σ3 ) + (σ1 + σ3 ) sin φ − 2 c cos φ = 0 .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
414 C HAPTER 8. P LASTICITY

Figure 8.31: Deduction of the expression for the Mohr-Coulomb criterion using Mohr’s
circle.

rs
ee
s gin
t d le En
Mohr-Coulomb criterion:
(8.63)
σ ) ≡ (σ1 − σ3 ) + (σ1 + σ3 ) sin φ − 2 c cos φ = 0
F (σ

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n

Remark 8.21. The equation


y ha

le
σ ) ≡ (σ1 − σ3 ) + (σ1 + σ3 ) sin φ − 2 c cos φ = 0 ,
F (σ
liv or ec
M

.A

which is linear in σ1 and σ3 , defines a plane in the principal stress


space that is restricted to the sector σ1 ≥ σ2 ≥ σ3 . Extension, taking
m

into account symmetry conditions, to the other five sectors (see Re-
d
uu

mark 8.7) defines six planes that constitute a pyramid of indefinite


e
X Th

er

length whose axis is the hydrostatic stress axis (see Figure 8.32). The
tin

distance from the origin


√ of the principal stress space to the vertex of
the pyramid is d = 3 c cot φ .
on

.O
C

Remark 8.22. The particularization φ = 0 and c = σe /2 reduces


the Mohr-Coulomb criterion to the Tresca criterion (see (8.59) and
(8.63)).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Yield Surfaces. Failure Criteria 415


3 c cot φ

rs
ee
s gin
Figure 8.32: Mohr-Coulomb criterion in the principal stress space.

t d le En

r
ba
ge ro or
eS m
ci
f

ra
Remark 8.23. In soil mechanics, the sign criterion of the normal
C d P cs
b
a
stresses is the opposite to the one used in continuum mechanics
i
(σ ≡ −σ , see Chapter 4) and, thus, σ1 ≡ −σ3 and σ3 ≡ −σ1 . Then,
an an n
y ha

the Mohr-Coulomb criterion in (8.63) becomes


le
liv or ec

σ ) ≡ (σ1 − σ3 ) − (σ1 + σ3 ) sin φ − 2 c cos φ .


F (σ
M

.A

The corresponding graphical representations are shown in Fig-


m

ures 8.33 and 8.34.


d
uu
e
X Th

er
tin
on

.O
C

Figure 8.33: Representation of the Mohr-Coulomb criterion using Mohr’s circle in three
dimensions and soil mechanics sign criterion.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
416 C HAPTER 8. P LASTICITY

rs
Figure 8.34: Mohr-Coulomb criterion in the principal stress space, using soil mechanics

ee
sign criterion.

s gin
t d le En

r
ba
ge ro or
eS m
Remark 8.24. Following certain algebraic operations, the Mohr-

ci
f
Coulomb criterion can be written in terms of the three stress invari-

ra
C d P cs
ants.
b
a
 
i
an an n

 
F σ
(σ ) ≡ F I , J , J
y ha

Mohr-Coulomb criterion: 1 2 3
le
liv or ec
M

.A
m

Remark 8.25. The Mohr-Coulomb criterion is especially adequate


uu
e

for cohesive-frictional materials (concrete, rocks and soils), which


X Th

er
tin

are known to exhibit considerably different uniaxial elastic limits


under tensile and compressive loadings.
on

.O
C

8.8.4 Drucker-Prager Criterion


The yield surface defined by the Drucker-Prager criterion is given by

  1/2
Drucker-Prager criterion: σ ) ≡ 3 ασm + J2
F (σ −β = 0 (8.64)

where

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Yield Surfaces. Failure Criteria 417

2 sin φ 6 c cos φ σ1 + σ2 + σ3 I1
α=√ ,β=√ and σm = = , (8.65)
3 (3 − sin φ ) 3 (3 − sin φ ) 3 3

being c and φ the cohesion and the internal friction angle, respectively, which are
considered to be material properties. Considering (8.16) and (8.18), the criterion
can be rewritten as
  1/2   
3 
σ ) ≡ αI1 + J2
F (σ − β = 3 ασoct + τoct − β = F I1 , J2 = 0 . (8.66)
2

rs
ee

Remark 8.26. The independence on the third stress invariant J3 es-

s gin
tablishes that, if a certain point in the stress space belongs to the
yield surface, all the other points with the same value of the stress

t d le En

invariants I1 and J2 also belong to this surface, independently of the


r
value of the third stress invariant J3 . Given that the constant values of

ba
ge ro or
eS m
these invariants correspond to points of the octahedral plane placed

ci
f
at a same distance from the hydrostatic stress axis (see Figure 8.6),

ra
C d P cs
b
it can be concluded that the yield surface is a surface of revolution

a
i
around this axis.
an an n

In addition, because the relation between σoct and τoct in (8.66)


y ha

is lineal, the surface is a conical surface whose axis is the hydro-


le
liv or ec

static stress axis (see Figure 8.5 and Figure 8.35). The distance from
M

.A

the origin
√ of the principal stress space to the vertex of the cone is
d = 3 c cot φ . It can be verified that the Drucker-Prager surface has
m

the Mohr-Coulomb surface with the same values of cohesion, c, and


d
uu

internal friction angle, φ , semi-inscribed in it.


e
X Th

er
tin
on

.O
C

Drucker-
Prager
Mohr-
Coulomb


3 c cot φ

Figure 8.35: Drucker-Prager criterion in the principal stress space.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
418 C HAPTER 8. P LASTICITY

Remark 8.27. The position of the vertex of the Drucker-Prager cone


in the positive side of the hydrostatic stress axis establishes a lim-
itation in the elastic behavior range for hydrostatic stress states in
tensile loading (while there is no limitation in the elastic limit for
the hydrostatic compression case). This situation, which also occurs
in the Mohr-Coulomb criterion, is typically observed in cohesive-
frictional materials (concrete, rocks and soils), for which these two
criteria are especially adequate.

rs
ee
s gin
Remark 8.28. In soil mechanics, where the sign criterion for the nor-

t d le En
mal stresses is inverted, the yield surface for the Drucker-Prager cri-
terion is as indicated in Figure 8.36.

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n

Remark 8.29. The particularization φ = 0 and c = σe /2 reduces


y ha

the Drucker-Prager criterion to the von Mises criterion (see (8.56),


le
liv or ec

(8.64) and (8.65)).


M

.A
m

d
uu
e
X Th

er
tin
on

.O

Drucker-
Prager
C

Mohr-
©

Coulomb

Figure 8.36: Drucker-Prager criterion in the principal stress space, using soil mechanics
sign criterion.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 419

P ROBLEMS

Problem 8.1 – Justify the shape the yield surface will have in the principal
stress space for each of the following cases:

a) f I12 = 0
b) f (J  2 ) = 0

rs
c) a I12 + b τoct
2 = c with a, b and c strictly positive

ee
s gin
t d le En
Solution

r
ba
ge ro or
a) In this case, there is a condition on the mean stress since

eS m
ci
f
I1 = σ1 + σ2 + σ3 = 3σm .

ra
C d P cs
b
a
i
Then, the yield surface is an octahedral plane whose distance to the origin is
an an n
y ha

imposed by the first stress invariant. However, because this invariant is squared,
there are two octahedral planes, one in each direction of the hydrostatic stress
le
liv or ec

axis.
M

.A
m

d
uu


e

3 σm
X Th

er
tin


3 σm
on

.O
C

hydrostatic stress axis


©

b) Here, the distance between a given stress state and an hydrostatic stress
state is imposed. So, the yield surface is a cylinder with circular section in the
octahedral planes,
3 2 √
J  2 = τoct =⇒ distance = 3 τoct .
2

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
420 C HAPTER 8. P LASTICITY


3 τoct

hydrostatic stress axis

rs
ee
s gin
c) The representation of a plane defined by a given point of the yield surface
and the hydrostatic stress axis is:

t d le En

r
ba
ge ro or
eS m
Point

ci
f

ra
C d P cs
b
a
i
an an n
y ha

= hydrostatic stress axis


le
liv or ec

Then, the relations


M

.A

√ ⎫ ⎧ √
√ 3 ⎬ ⎨ I1 = 3 x
m

d=x= 3 σoct = I1
d

=⇒ R
uu

√ 3
⎭ ⎩ τoct = √
e

R = y = 3 τoct
X Th

3
er
tin

are deduced and replacing these values in the given expression of the yield sur-
on

.O

face results in
C

⎛ ⎞2 ⎛ ⎞2
©

by2 ⎜ x ⎟ ⎜ y ⎟
a I1 + b τoct = c =⇒ 3ax +
2 2 2
= c =⇒ ⎜ ⎟ ⎜
⎝ c ⎠ + ⎝ 3c ⎠ = 1 .

3
3a b

This is the mathematical description of an ellipse in the x − y plane previously


defined. In addition, since the third stress invariant does not intervene in the
definition of the yield surface, the hydrostatic stress axis is an axis of radial
symmetry and, thus, the rotation of the ellipse about the x-axis (≡ hydrostatic
stress axis) defines the final surface.
In conclusion, if the axes considered are the axes x (≡ hydrostatic stress axis), y
and z, the yield surface is defined by

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 421

⎛ ⎞2 ⎛ ⎞2 ⎛ ⎞2
⎜ x ⎟ ⎜ y ⎟ ⎜ z ⎟
⎜ c ⎟ +⎜ ⎟ ⎜ ⎟
⎝ ⎠ ⎝ 3c ⎠ + ⎝ 3c ⎠ = 1 .
3a b b

rs
ee
hydrostatic stress axis

s gin
t d le En

r
ba
ge ro or
eS m
Problem 8.2 – Graphically determine, indicating the most significant values,

ci
f

ra
the cohesion and internal friction angle of an elastoplastic material that follows
C d P cs
b
a
the Mohr-Coulomb yield criterion using the following information:
i
an an n

1) In an uniaxial tensile stress state (σ1 = σ , σ2 = σ3 = 0), the material


y ha

plasticizes at σ = σA .
le
liv or ec

2) In a triaxial isotensile test of the same material (σ1 = σ2 = σ3 = σ ), it


M

.A

plasticizes at σ = σB .
m

d
uu
e
X Th

er

Solution
tin

In the uniaxial tensile stress state, the Mohr’s circle will cross the origin and the
on

.O

value σ = σA in the horizontal axis. However, for the triaxial isotensile stress
C

state, the Mohr’s circle will degenerate to a point in this axis, σ = σB . Thus, the
©

following graph is plotted

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
422 C HAPTER 8. P LASTICITY

which allows establishing the relations


c σA /2
tan φ = and sin φ = .
σB σB − σA /2
Finally, the cohesion and internal friction angle are

σA /2
φ = arcsin and c = σB tan φ .
σB − σA /2

rs
ee
s gin
Problem 8.3 – The following properties of a certain material have been exper-
imentally determined:

t d le En
1) In a hydrostatic compressive regime, the material never plasticizes.

r
ba
ge ro or
2) In a hydrostatic tensile regime, the virgin material plasticizes for a value of

eS m
ci
the mean stress σm = σ ∗ .
f

ra
C d P cs
b
a
3) In an uniaxial tensile regime, the virgin material plasticizes for a tensile
i
stress value σu .
an an n
y ha

4) In other cases, plasticization occurs when the norm of the deviatoric


le
liv or ec

stresses varies linearly with the mean stress,


  √
M

σ  = σ  : σ  = a σm + b .
.A
m

Plot the yield surface, indicating the most significant values, and calculate the
d

values a and b in terms of σ ∗ and σu .


uu
e
X Th

er
tin
on

.O

Solution
C

Property 1) and 2) indicate that the


©

yield surface is closed in the tensile


part of the hydrostatic stress axis but
open in the compressive part. In addi-
tion, property 3) indicates that the oc-
tahedral plane that contains the origin
will have the shape shown in the figure
to the right. Since property 4) indicates
that the deviatoric stresses vary linearly
with the mean stress (as is the case for
the Drucker-Prager criterion), then the
yield surface is necessarily a right cir-
cular cone whose axis is the hydrostatic

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 423

stress axis and whose vertex is in the tensile part of this axis:

√ √
3 σm = 3σ∗

hydrostatic
stress axis

rs
ee
s gin

σ  | = σ  : σ  = a σm + b
To calculate the values of a and b, the yield criterion |σ

t d le En
is applied on the vertex of the cone, which corresponds with the hydrostatic

r
tensile case and, thus, has no deviatoric stresses,

ba
ge ro or
eS m



ci
 + b = 0 =⇒ aσ ∗ + b = 0 .
σ | = 0 =⇒ aσm 
|σ f [1]

ra
C d P cs
∗ σm =σ

b
a
i
an an n

The procedure is repeated for the uniaxial tensile case, whose deviatoric stresses
y ha

are now
le
liv or ec

⎡ ⎤ ⎡ ⎤
σu 0 0 2 0 0
M

.A

not ⎢ ⎥ 1 not σu ⎢ ⎥
σ ≡ ⎣ 0 0 0 ⎦ and σ sph = σu 1 =⇒ σ  ≡ ⎣ 0 −1 0 ⎦ .
m

3 3
d

0 0 0 0 0 −1
uu
e
X Th

er

σ  | = a σm + b produces
Then, applying the yield criterion |σ
tin

   
on

.O

2 2 1
σ |
|σ = σu =⇒ σu = a σu + b . [2]
C

3 3 3
©

Equations [1] and [2] allow determining the desired values of a and b as
 
2 2
σu σu σ ∗
3 3
a= σ and b=− σ .
u u
−σ∗ −σ∗
3 3

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
424 C HAPTER 8. P LASTICITY

Problem 8.4 – The metallic component PQRS has a thickness “e” and is com-
posed of two different materials, (1) and (2), considered to be perfect elasto-
plastic materials. The component is subjected to a pure shear test by means of
the machine shown in Figure A, such that the uniform stress and strain states
produced are
δ
εx = εy = εz = 0 , γxz = γyz = 0 , γxy = γ = ,
h
σx = σy = σz = 0 , τxz = τyz = 0 and τxy = τ
= 0 .
When a component exclusively composed of one of the materials is tested sep-

rs
arately, a τ − γ curve of the type shown in Figure B is obtained for both mate-

ee
rials. Determine:

s gin
a) The elastic limit that will be obtained in separate uniaxial tensile tests of
each material, assuming they follow the von Mises criterion.

t d le En
When the component composed of the two materials is tested, the P − δ curve

r
ba
ge ro or
shown in Figure C is obtained. Determine:

eS m
ci
f
b) The values of the elastic load and displacement, Pe and δe .

ra
C d P cs
b
c) The values of the plastic load and displacement, Pp and δp .

a
i
an an n

d) The coordinates P − δ of points C and D in Figure C.


y ha

le
liv or ec

rigid frame
M

.A
m

d
uu
e
X Th

er
tin
on

.O

Figure A
C

H YPOTHESES :
Material (1)
G = G and τe = τ ∗
Material (2)
G = G and τe = 2 τ ∗

Figure B Figure C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 425

Solution
a) In an uniaxial state of stress, plasticization according to the von Mises crite-
rion is known to begin when (see Example 8.2)

σ̄ = σe ,

where σ̄ is the effective stress and σe is the elastic limit. In addition, the follow-
ing relations seen in this chapter, are known to hold.
1 1  2
σ̄ = (3 J2 ) 2 J2 = Tr σ

rs
2
1

ee
σ  = σ − σ sph σ sph = σm 1 σm = σ)
Tr (σ
3

s gin
For this problem in particular,

t d le En
⎡ ⎤
0 τ 0

r
ba
not ⎢ ⎥

ge ro or
eS m
σ ≡⎣ τ 0 0 ⎦ ,

ci
f

ra
0 0 0
C d P cs
b
a
i
so σm = 0 and, therefore, σ sph = 0, leading to σ  = σ . Then,
an an n
y ha

⎡ ⎤
le
τ2 0 0
liv or ec

 2 not ⎢ ⎥ √
σ ≡ ⎣ 0 τ2 0 ⎦ =⇒ J2 = τ 2 =⇒ σ=
σ̄ 3τ .
M

.A

0 0 0
m

d
uu

Considering that material (1) plasticizes when τe = τ ∗ and material (2), when
e
X Th

τe = 2τ ∗ , then
er
tin


Material 1 =⇒ σe = 3 τ ∗ ,
on

.O


Material 2 =⇒ σe = 2 3 τ ∗ .
C

b) The elastic load Pe and the elastic displacement δe determine the end of the
elastic regime in the component. The statement of the problem indicates that
when the materials are tested separately, the τ − γ curve in Figure B is obtained,
where τe = τ ∗ in material (1) and τe = 2τ ∗ in material (2). It is also known that G
is the same in both materials, that is, they have the same slope in their respective
τ − γ curves.
Now, to determine the combined behavior of these materials in the metallic
component, one can assume that the behavior will be elastic in this component
as long as both materials are in their corresponding elastic domain. Therefore,

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
426 C HAPTER 8. P LASTICITY

since the elastic interval of material (1) is smaller, then this material will define
the elastic domain of the whole component (up to point A in Figure C).

rs
ee
s gin
To obtain the value of the elastic force, equilibrium of forces is imposed for the

t d le En
force Pe and the stresses each material has at point A. Note that equilibrium is im-
posed on forces, therefore, stresses must be multiplied by the surface on which

r
ba
ge ro or
they act, considering the magnitude perpendicular to the plane of the paper as

eS m
ci
the unit value.
f

ra
C d P cs
b
h A h A h ∗ h ∗

a
Pe = τ + τ = τ + τ =⇒ Pe = h τ ∗
i
2 1 2 2
an an n

2 2
y ha

The elastic displacement is obtained imposing kinematic compatibility of the


le
liv or ec

two materials,
τ∗
M

.A

δe = γ1A h = γ2A h =⇒ δe = h .
G
m

d
uu
e

c) To obtain the plastic values Pp and δ p one must take into account that, at
X Th

er
tin

point A, material (1) begins to plasticize, while material (2) initiates plasticiza-
tion at point B. Therefore, the behavior of the complete component will be per-
on

.O

fectly plastic starting at point B, but elastoplastic between points A and B. To


determine the coordinates of point B, the same procedure as before is used. Plot-
C

ting the τ − γ curves of each separate material up to point B results now in


©

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 427

and, imposing the equilibrium and compatibility equations, yields the values of
Pp and δ p .

h B h B h ∗ h ∗⎪ ⎪ 3
Pp = τ1 + τ2 = τ + 2τ ⎬ Pp = τ ∗ h and
2 2 2 2 =⇒ 2
2τ ∗ ⎪
⎪ τ ∗h
δ p = γ1B h = γ2B h = h ⎭ δp = 2 = 2δe .
G G

rs
d) The coordinates of points A and B have already been obtained. The statement
of the problem gives the value of point B , which corresponds to a deformation

ee
of 3δe when the plastic load Pp is maintained constant (perfectly plastic regime).

s gin
Consider first the material (1). Unloading takes place starting at B and, accord-
ing to the information given, this material plasticizes when it reaches a value

t d le En
of −τ ∗ . The slope of the curve is still the value of the material parameter G

r
since this is independent of the material being under loading or unloading con-

ba
ge ro or
eS m
ditions. Thus, to determine point C it is enough to draw a straight line that crosses

ci
point B and is parallel to OA, until the value −τ ∗ is reached.
f

ra
C d P cs
The same occurs in the case of material (2), with the difference that when the
b
a
line parallel to OA is drawn to cross point B , this line must be extended to the
i
an an n

value −2τ ∗ (which corresponds to point D).


y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Then, the load and displacement values at point B are


3τ ∗ h 3
δB = 3δA = = 3δe and PB = PB = τ ∗ h .
G 2

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
428 C HAPTER 8. P LASTICITY

To obtain the load and displacement values at point C, the equilibrium and com-
patibility equations are imposed. Taking into account the τ and γ values obtained
at point C in the curves above yields

∗h ⎪
h C h C h h τ ⎪
⎪ τ ∗h
PC = τ1 + τ2 = (−τ ∗ ) + (0) = − ⎬ PC = − and
2 2 2 2 2 =⇒ 2
 ∗
τ ⎪
⎪ τ ∗h
δC = γ1C h = h ⎪
⎭ δC = = δe .
G G

rs
Repeating the procedure for point D results in

ee
∗h ⎪
h D h D h h 3τ ⎪ 3τ ∗ h

s gin
PD = τ1 + τ2 = (−τ ∗ ) + (−2τ ∗ ) = − ⎪
⎬ PD = − and
2 2 2 2 2 ⇒ 2
 ∗

t d le En
τ ⎪
⎪ τ ∗h
δD = γ2D h = − h ⎪
⎭ δD = − = −δe .

r
G G

ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n

Problem 8.5 – Consider the solid cylinder shown in Figure A, which is fully
y ha

fixed at its base and has a torsional moment M applied on its top end. The
le
cylinder is composed of two materials, (1) and (2), which have an elastoplas-
liv or ec

tic tangent stress-strain behavior, as shown in Figure B. Assume the following


M

.A

displacement field in cylindrical coordinates (Coulomb torque),


, -T
m

θ
d

not T
u (r, θ , z) ≡ [ur , uθ , uz ] = 0 , rz , 0 ,
uu
e

h
X Th

er
tin

where φ is the rotation of the section at the free end of the cylinder. Assuming
on

.O

infinitesimal strains, determine:


C

a) The strain and stress tensors, ε and σ , in cylindrical coordinates and elas-
©

tic regime. Plot, indicating the most significant values, the σrr − r and
τθ z − r curves for a cross-section of the cylinder at height z. Schematically
represent the stress distribution of τθ z in this cross-section.
b) The value of φ = φe (see Figure C) for which plasticization begins in at least
one point of the cylinder, indicating where it begins and the corresponding
value of the moment M = Me .
NOTE: M = r τθ z dS
S
c) The minimum value of φ = φ1 for which material (1) has totally plasticized
and the corresponding value of M = M1 (see Figure C). Schematically rep-
resent the stress distribution in a cross-section at this instant.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 429

d) The minimum value of φ = φ2 for which material (2) has totally plasticized
and the corresponding value of M = M2 (see Figure C). Schematically rep-
resent the stress distribution in a cross-section at this instant.
e) The asymptotic value of M = M p (= plastic moment) corresponding to the
plasticization of the complete cross-section. Schematically represent the
stress distribution in a cross-section at this instant.

rs
ee
s gin
t d le En
Figure B

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

Figure A
.A

Figure C
m

H YPOTHESES :
d

Material (1): G = G and τe = τ ∗ .


uu
e

Material (2): G = G and τe = 2 τ ∗ .


X Th

er
tin
on

.O
C

Solution
©

a) The infinitesimal strain tensor is calculated directly from the given displace-
ment field, both in cylindrical coordinates,

⎡ ⎤
0 0 0
⎢ φ r ⎥
not ⎢ ⎥
ε ≡⎢ 0 0 ⎥ .
⎣ φr
2h ⎦
0 0
2h

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
430 C HAPTER 8. P LASTICITY

To compute the stress tensor, the constitutive equation of an isotropic elastic


material is used. Note that the two materials composing the cylinder have the
same parameter G, then

σ = λ Tr (εε ) 1 + 2μεε and


.
Tr (εε ) = 0
=⇒ σ = 2Gεε .
μ =G

The stress tensor results in

rs
⎡ ⎤

ee
0 00
⎢ ⎥

s gin
not ⎢ Gφ r ⎥
σ ≡⎢ 0 0 ⎥ .
⎣ h ⎦

t d le En
Gφ r
0 0
h

r
ba
ge ro or
eS m
ci
Plotting the σrr and τθ z components of the stress tensor in terms of the radius r
f

ra
C d P cs
yields:
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin

The stresses are linear and do not depend on the z-coordinate of the cross-section
considered. Thus, the distribution of stresses in any cross-section (z = const.) of
on

.O

the cylinder is:


C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 431

b) Given the stress distribution τ = (Gφ r/h) ≤ φ ≤ φe , the moment acting on


the cylinder is
2π R   R
Gφ r Gφ r3 πGR4
M= r τ (r) dS = r r dr dθ = 2π dr = φ. [1]
h h 2h
S 0 0 0

This is the relation between the moment and the rotation angle (M − φ ) at the
free end of the cylinder when the two materials behave elastically.
Material (1) starts to plasticize first at τe = τ ∗ , since material (2) plasticizes at a
higher stress, τe = 2τ ∗ . In addition, the external surface of the cylinder (r = R)

rs
suffers larger stresses, and this surface is composed of material (1). Therefore,

ee
plasticization will initiate when

s gin
 τ ∗h
 Gφe R

t d le En
τ = τ∗ =⇒ = τ∗ =⇒ φe =
r=R; φ =φe h GR

r
ba
ge ro or
eS m
is satisfied. This is the value of the rotation angle at the free end of the cylinder

ci
f
required for plasticization to initiate in the exterior material points of the cylinder

ra
C d P cs
(material (1)). The corresponding moment is obtained by replacing φe in [1],
b
a
i
an an n

πτ ∗ R3
y ha

πGR4
Me = M (φe ) = φe =⇒ Me = .
le
2h 2
liv or ec
M

.A

c) If the material were elastic, the slope of the stresses τ would increase with φ
m

(remaining, though, linear with r), but since the material is now elastoplastic,
uu

stresses cannot exceed the value τe , which corresponds to the onset of plasticity.
e
X Th

Then, the limit value is obtained for τe = τ ∗ when φ = φ1 for (R/2) ≤ r ≤ R.


er
tin

That is, material (1) has a perfectly plastic distribution of stresses while material
on

.O

(2) remains elastic.


C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
432 C HAPTER 8. P LASTICITY

The following condition is imposed to compute this rotation φ1 .


 2τ ∗ h
 Gφ1 R
τ = τ∗ =⇒ = τ∗ =⇒ φ1 = .
r=R/2; φ =φ1 2h GR

This is the minimum value of the rotation angle at the free end of the cylinder
required for material (1) to be completely plasticized.
In order to compute the corresponding moment, relation [1] between M and
φ is no longer valid here because material (1) behaves elastoplastically while
material (2) behaves completely elastically. The moment acting on the cylinder

rs
is now

ee
2π R 2π R/2  

s gin
∗ Gφ1 r
M1 = rτ r dr dθ + r r dr dθ =
h

t d le En
0 R/2 0 0
R R/2
φ1

r
∗ 31 ∗ 3

ba
= 2πτ r dr + 2πG
2
r3 dr =⇒ M1 = πτ R .

ge ro or
eS m
h 48

ci
R/2 0
f

ra
C d P cs
b
a
i
an an n

d) Material (2) starts plasticizing for τe = 2τ ∗ , which does not correspond with
y ha

the end of plasticization in material (1) at τe = τ ∗ . Then, the stress distribution


le
for φ = φ2 (onset of plasticization in material (2)) is
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

The following condition is imposed to obtain the value of the rotation angle.
 4τ ∗ h
 Gφ2 R
τ = 2τ ∗ =⇒ = 2τ ∗ =⇒ φ2 = .
r=R/2; φ =φ2 2h GR

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 433

The corresponding moment is


2π R 2π R/2  
∗ Gφ2 r 17 ∗ 3
M2 = r τ r dr dθ + r r dr dθ =⇒ M2 = πτ R .
h 24
0 R/2 0 0

e) The asymptotic value of M (M p ) corresponds to the total plasticization of the


cylinder. The stress distribution in this case is:

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec

Through integration, the corresponding moment is obtained,


M

.A

2π R 2π R/2
3
m

Mp = r τ ∗ r dr dθ + r 2τ ∗ r dr dθ =⇒ M p = πτ ∗ R3 .
d

4
uu
e

0 R/2 0 0
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
434 C HAPTER 8. P LASTICITY

E XERCISES

8.1 – Formulate in terms of the stress invariants I1 , J2 and J3 the equation of
the yield surface that, in the principal stress space, is a spheroid (ellipsoid of
revolution) with semi-axes a and b.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
Intersection with octahedral

ci
plane at (0, 0, 0)
f

ra
C d P cs
b
a
i
an an n
y ha

8.2 – An elastoplastic material is subjected to a pure shear test (I) and an uni-
le
axial tensile test (II). Plasticization occurs, respectively, at τ = a and σ = b.
liv or ec

Determine the values of the cohesion and internal friction angle assuming a
M

.A

Mohr-Coulomb yield criterion.


m

d
uu
e
X Th

er
tin
on

.O
C

8.3 – A component ABCD of a perfectly elastoplastic material is tested in the


machine illustrated in Figure A. The action-response curve (P − δ ) obtained is
shown in Figure B. An uniaxial stress-strain state is assumed such that
δ
εx = y and εy = εz = γxy = γxz = γyz = 0 ,
hL
σx
= 0 and σy = σz = τxy = τxz = τyz = 0 .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 435

Determine the following values, indicated in the curve of Figure B:


a) The elastic load Pe and the corresponding displacement δe .
b) The ultimate plastic loads for tensile and compressive loadings, Pp and
Pq , respectively.
c) The values of P and δ at points (1) and (2).

rs
ee
s gin
t d le En
Infinitely rigid
material

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

Figure A
le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure B
Additional hypotheses:
1) Young’s modulus, E, and Poisson’s coefficient, ν.
2) Elastic limit, σe .
3) Thickness of the component, b.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
436 C HAPTER 8. P LASTICITY

8.4 – The truss structure OA, OB and OC is composed of concrete, which is


assumed to behave as a perfectly elastoplastic material with a tensile elastic
limit σe and a compressive elastic limit 10 σe . An increasing vertical load P is
applied at point O, starting at P = 0, until a vertical displacement δ = 20 σe L/E
is reached at this point. Then, the load is decreased back to P = 0 .
a) Draw the P − δ diagram of the process, indicating the most significant
values and the state of plasticization of the bars at each instant.
b) Calculate the displacement value at point O at the end of the process.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

8.5 – Consider a solid sphere with radius R1 encased inside a spherical shell
d
uu

with interior radius R1 and exterior radius R2 . The sphere and the shell are
e
X Th

composed of the same material and are initially in contact without exerting any
er
tin

pressure on each other. At a certain moment, the interior sphere is heated up to


a temperature increment Δ θ .
on

.O

Determine:
C

a) The value of the exterior pressure required


on the shell for said shell to keep a constant
value (infinitesimal strain hypothesis).
b) The displacement, strain and stress fields in
both the sphere and the shell under these
conditions.
c) The minimum value of Δ θ for which plas-
ticization initiates in some point, assuming
the aforementioned conditions and consid-
ering a von Mises criterion.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 437

Additional hypotheses:
1) Material properties:
− Young’s modulus, E, and Poisson’s coefficient, ν = 0.
− Thermal constant, α.
− Yield stress, σy .
− Radii, R1 = 1 and R2 = 3.
2) The body forces are negligible.
3) The displacement and stress fields of a spherical shell with interior radius Ri
and exterior radius Re subjected to an interior pressure Pi and an exterior

rs
pressure Pe are, for ν = 0:
⎡ ⎤

ee
ur (r)
Pi R3 − Pe R3e Pi − Pe R3i R3e

s gin
⎢ ⎥ C1
u=⎣ 0 ⎦ ur = Cr + 2 ; C = i 3  ; C1 =
r E Re − R3i 2E R3e − R3i

t d le En
0
⎡ ⎤

r
ba
σrr 0    

ge ro or
0

eS m
⎢ ⎥ 2C1 C1

ci
σ = ⎣ 0 σθ θ 0 ⎦ σrr = E C − 3 ; σθ θ = σφ φ = E C + 3
f

ra
r r
C d P cs
b
a
0 0 σφ φ
i
an an n
y ha

le
liv or ec

8.6 – Consider a solid sphere with radius R1 and composed of material (1), en-
M

.A

cased inside a spherical shell with interior radius R1 , exterior radius R2 and
composed of material (2). The sphere and the shell are initially in contact with-
m

out exerting any pressure on each other. An exterior pressure P is applied simul-
uu

taneously with a temperature increment Δ θ .


e
X Th

er
tin

a) Determine the possible values of Δ θ and P (positive or negative) for which


the contact (without exerting any pressure) between the sphere and the shell
on

.O

is maintained. Plot the corresponding P − Δ θ curve.


C

b) Obtain the stress state of the shell and the sphere for these values.
©

c) Under these conditions, compute, for each value of the pressure P, the value
of Δ θ ∗ for which plasticization initiates at some point of the sphere or the
shell, according to the von Mises and Mohr-Coulomb criteria. Plot the cor-
responding P − Δ θ ∗ curves (interaction graphs).

Additional hypotheses:
1) Material properties:
− Young’s moduli, E (1) = E (2) = E, and Poisson’s coefficients, ν (1) = ν (2) = 0.
− Thermal constants, α (1) = 2α and α (2) = α.
(1) (2)
− Yield stresses, σy = σy = σy .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
438 C HAPTER 8. P LASTICITY

− Cohesion values, C(1) = C(2) = C, and internal friction angles, φ (1) =


φ (2) = 30◦ .
− Radii, R1 = 1 and R2 = 2.
2) The displacement and stress fields of a spherical shell with interior radius Ri
and exterior radius Re subjected to an interior pressure Pi and an exterior
pressure Pe are, for ν = 0:
⎡ ⎤
ur (r)
⎢ ⎥ C1 Pi R3 − Pe R3e Pi − Pe R3i R3e
u=⎣ 0 ⎦ ur = Cr + 2 ; C = i 3  ; C1 =
r E Re − R3i 2E R3e − R3i

rs
0
⎡ ⎤

ee
σrr 0 0    

s gin
⎢ ⎥ 2C1 C1
σ = ⎣ 0 σθ θ 0 ⎦ σrr = E C − 3 ; σθ θ = σφ φ = E C + 3
r r

t d le En
0 0 σφ φ

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
8.7 – A cylinder of radius R and height h is subjected to an exterior load P and
b
a
a uniform temperature increment Δ θ .
i
an an n
y ha

a) Determine the displacement, strain and


le
tensor fields in terms of the integration
liv or ec

constants.
M

.A

b) Determine the integration constants and


the corresponding displacement, strain
m

and tensor fields.


uu
e

c) Given p = p∗ > 0, determine the corre-


X Th

er

sponding value of Δ θ ∗ such that there


tin

are no horizontal displacements.


on

.O

d) Under the conditions described in c),


determine the value of p∗ for which the
C

cylinder begins to plasticize according


to the Mohr-Coulomb criterion.

Additional hypotheses:
1) Material properties:
− Cohesion value, C, and internal friction angle, φ = 30◦ .
− Thermal constant, β .
− Lamé parameters, λ = μ.
2) The body forces are negligible.
3) The friction between the cylinder and the ground can be neglected.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
CH.9. CONSTITUTIVE
EQUATIONS IN FLUIDS
Multimedia Course on Continuum Mechanics
Overview
 Introduction
 Fluid Mechanics Lecture 1
 What is a Fluid?
 Pressure and Pascal´s Law Lecture 3
 Constitutive Equations in Fluids
Lecture 2
 Fluid Models
 Newtonian Fluids
 Constitutive Equations of Newtonian Fluids Lecture 4
 Relationship between Thermodynamic and Mean Pressures
 Components of the Constitutive Equation Lecture 5
 Stress, Dissipative and Recoverable Power
 Dissipative and Recoverable Powers Lecture 6
 Thermodynamic Considerations
 Limitations in the Viscosity Values

2
9.1 Introduction
Ch.9. Constitutive Equations in Fluids

3
What is a fluid?
Fluids can be classified into:
 Ideal (inviscid) fluids:
 Also named perfect fluid.
 Only resists normal, compressive stresses (pressure).
 No resistance is encountered as the fluid moves.

 Real (viscous) fluids:


 Viscous in nature and can be subjected to low
levels of shear stress.
 Certain amount of resistance is always offered
by these fluids as they move.

5
9.2 Pressure and Pascal’s Law
Ch.9. Constitutive Equations in Fluids

6
Pascal´s Law
 Pascal’s Law:
In a confined fluid at rest, pressure acts equally in all
directions at a given point.

7
Consequences of Pascal´s Law
 In fluid at rest:
 there are no shear stresses
 only normal forces due to pressure are present.

 The stress in a fluid at rest is isotropic and must be of the form:

σ = − p0 1
− p0δ ij i, j ∈ {1, 2,3}
σ ij =

Where p0 is the hydrostatic pressure.

8
Pressure Concepts
 Hydrostatic pressure, p0 : normal compressive stress exerted on a
fluid in equilibrium.

 Mean pressure, p : minus the mean stress.


REMARK
1
p = −σ m = − Tr ( σ ) Tr ( σ ) is an invariant,
3 thus, so are σ m and p .

 Thermodynamic pressure, p : Pressure variable used in the


constitutive equations . It is related to density and temperature
through the kinetic equation of state. REMARK
F ( ρ , p, θ ) = 0 In a fluid at rest,
p0= p= p

9
Pressure Concepts
 Barotropic fluid: pressure depends only on density.

F ( ρ , p ) 0=
= p f (ρ)

 Incompressible fluid: particular case of a barotropic fluid in


which density is constant.

F ( ρ , p,θ ) ≡ F ( ρ ) = ρ − k = 0 ρ = k = const.

10
9.3 Constitutive Equations
Ch.9. Constitutive Equations in Fluids

11
Reminder – Governing Eqns.
 Governing equations of the thermo-mechanical problem:
Conservation of Mass.
ρ + ρ ∇ ⋅ v = 0 Continuity Equation.
1 eqn.

Linear Momentum Balance.


∇ ⋅ σ + ρ b = ρ v 3 eqns.
Cauchy’s Motion Equation.
Angular Momentum Balance. Symmetry
8 PDE + σ =σT of Cauchy Stress Tensor.
3 eqns.
2 restrictions
ρ u σ : d + ρ r − ∇ ⋅ q
= Energy Balance. First Law of
1 eqn.
Thermodynamics.

− ρ ( u − θ s ) + σ : d ≥ 0 Clausius-Planck
Inequality. Second Law of
1 2 restrictions
− 2 q ⋅ ∇θ ≥ 0 Heat flux Thermodynamics.
ρθ Inequality.

 19 scalar unknowns: ρ , v , σ , u , q , θ , s .

12
Reminder – Constitutive Eqns.
 Constitutive equations of the thermo-mechanical problem:
Thermo-Mechanical
σ = σ ( v, θ , ζ ) Constitutive Equations.
6 eqns.

Entropy
s = s ( v, θ , ζ ) Constitutive Equation. 1 eqn.
(19+p) PDE +
(19+p) unknowns Thermal Constitutive Equation. Fourier’s
q=q (θ ) =− K ∇θ Law of Conduction. 3 eqns.

u = f ( ρ , v, θ , ζ ) Caloric
State Equations. (1+p) eqns.
Fi ( ρ ,θ=
,ζ ) 0 i ∈ {1, 2,..., p} Kinetic

set of new thermodynamic


variables: ζ = {ζ 1 , ζ 2 ,..., ζ p } .
 The mechanical and thermal problem can be uncoupled if the temperature
distribution is known a priori or does not intervene in the constitutive eqns. and
if the constitutive eqns. involved do not introduce new thermodynamic variables.

13
Constitutive Equations
 Constitutive equations
 Together with the remaining governing equations, they are used to
solve the thermo/mechanical problem.

 In fluid mechanics, these are grouped into:


Thermo-mechanical constitutive equations Caloric equation of state
σ = − p 1 + f ( d, ρ , θ ) u = g ( ρ ,θ )

σ ij =− pδ ij + f ij ( d, ρ , θ ) i, j ∈ 1, 2,3

Entropy constitutive equation Kinetic equation of state


s = s ( d, ρ , θ ) F ( ρ , p, θ ) = 0
q =−k ⋅ ∇θ
 REMARK
Fourier’s Law  ∂θ d ( v ) = ∇s v
q
 i =
− k i, j ∈ 1, 2,3
 ∂xi
14
Viscous Fluid Models
 General form of the thermo-mechanical constitutive equations:

− p 1 + f ( d, ρ , θ )
σ=
− pδ ij + f ij ( d, ρ , θ ) i, j ∈ {1, 2,3}
σ ij =

 Depending on the nature of f ( d, ρ , θ ) , fluids are classified into :


1. Perfect fluid: f ( d, ρ , θ ) =⇒
0 σ=− p1
2. Newtonian fluid: f is a linear function of the strain rate
3. Stokesian fluid: f is a non-linear function of its arguments

15
9.4. Newtonian Fluids
Ch.9. Constitutive Equations in Fluids

16
Constitutive Equations of Newtonian
Fluids
 Mechanic constitutive equations:
σ= − p1 + C : d
σ ij =− pδ ij + Cijkl d kl i, j ∈ {1, 2,3}
where C is the 4th-order constant (viscous) constitutive tensor.

C = λ1 ⊗ 1 + 2 µ I
 Assuming:
Cijkl =λδ ijδ kl + µ (δ ik δ jl + δ ilδ jk )
 an isotropic medium
 the stress tensor is symmetrical i, j , k , l ∈ {1, 2,3}
 Substitution of C into the constitutive equation gives:
σ= − p1 + λTr ( d ) 1 + 2 µ d REMARK
σ ij =− pδ ij + λ dllδ ij + 2 µ dij i, j ∈ {1, 2,3} λ and µ are not necessarily constant.
Both are a function of ρ and θ .

17
Relationship between Thermodynamic
and Mean Pressures
 Taking the mechanic constitutive equation,
σ ij =
− pδ ij + λ dllδ ij + 2 µ dij i, j ∈ {1, 2,3}

 Setting i=j, summing over the repeated index, and noting that
δ ii = 3 , we obtain
−3 p + ( 3λ + 2µ ) dll =
σii = 1
−3 p ( p = − σ ii )
 3
−3 p Tr (d )

bulk viscosity κ
2
p = p + (λ + µ )Tr ( d ) = p + κ Tr ( d ) κ= λ + µ
2
3 3

18
Relationship between Thermodynamic
and Mean Pressures
 Considering the continuity equation,
dρ 1 dρ
+ ρ∇ ⋅ v = 0 ∇⋅v = −
dt ρ dt
And the relationship
p= p + κ Tr ( d )

∂v i
Tr (d ) = d ii = =∇⋅v
∂xi

REMARK κ dρ
For a fluid at rest, v= 0 p= p= p0 p= p + κ ∇ ⋅ v= p −

ρ dt
For an incompressible fluid,
= 0= p p
dt
2
For a fluid with , κ
=0 λ= − µ p=p
Stokes' 3
condition

19
9.5 Components of the
Constitutive Equations
Ch.9. Constitutive Equations in Fluids

20
Components of the Constitutive
Equation
 Given the Cauchy stress tensor, the following may be defined:
σ= − p1 + λTr ( d ) 1 + 2 µ d = σ σ sph + σ′
= − p1
 SPHERICAL PART – mean pressure p= p − κ ∇ ⋅ v= p − κ Tr ( d )

 DEVIATORIC PART

− p1 + λTr ( d ) 1 + 2 µ d =
− p1 + σ′ σ′ =( p − p ) 1 + λTr ( d ) 1 + 2 µ d
p= p − κ Tr ( d ) 2
2 κ= λ + µ
σ =−(λ + µ ) Tr ( d ) 1 + λTr ( d ) 1 + 2 µ d
′ 3
3
1 deviatoric part of the rate of
σ′ =2 µ (d − Tr ( d ) 1) = 2 µ d′ strain tensor
 3 
=d′
21
Components of the Constitutive
Equation
 Given the Cauchy stress tensor, the following may be defined:
 SPHERICAL PART – mean pressure p

p= p − κ ∇ ⋅ v= p − κ Tr ( d ) p
−κ

 DEVIATORIC PART – deviator stress tensor σ ij′ Tr ( d )

σ ′ = 2 µ d′

 The stress tensor is then dij′
1
σ =Tr ( σ ) 1 + σ′ = − p 1 + σ′ REMARK
3
= −3 p from the definition Note that κ is not a
of mean pressure function of d, while
µ = µ (d ).
22
9.6 Stress, Dissipative and
Recoverable Powers
Ch.9. Constitutive Equations in Fluids

23
Reminder – Stress Power
 Mechanical Energy Balance:

d 1 2
Pe ( t ) = ∫ ρ b ⋅ v dV + ∫ t ⋅ v dS = ∫ ρ v dV + ∫ σ : d dV
V ∂V dt Vt ≡V 2 V

external mechanical power kinetic energy stress power


entering the medium
d
Pe ( t )
= K ( t ) + Pσ
dt
REMARK
The stress power is the mechanical power entering the system which is not spent
in changing the kinetic energy. It can be interpreted as the work per unit of time
done by the stress in the deformation process of the medium.
A rigid solid will have zero stress power.

24
Dissipative and Recoverable Powers
 Stress Power = ∫ σ : d dV
1
V =d Tr (d)1 + d′
3
σ= − p1 + σ′
1 
(
σ : d =− p1 + σ ′ )  ( )
: Tr d 1 + d ′ =
3 
=3 = Tr =( d′ ) 0
1 1
=− p Tr ( d ) 1 : 1 + σ′ : d′ − p1 : d′ + Tr ( d ) σ′ : 1 =
3 3 = Tr =( σ′ ) 0
− p Tr ( d ) + σ′ : d′
=

σ ′ = 2 µ d′ − pTr ( d ) + κ Tr 2 ( d ) + 2 µ d′ : d′
σ :d =
p= p − κ Tr ( d )
RECOVERABLE STRESS DISSIPATIVE STRESS
POWER, WR. POWER, 2WD.

25
Dissipative and Recoverable Parts of
the Cauchy Stress Tensor
 Associated to the concepts of recoverable and dissipative
powers, the Cauchy stress tensor is split into:

− p 1 + λTr ( d ) 1 + 2 µ d
σ=
RECOVERABLE DISSIPATIVE
PART, σ R . PART, σ D .

 And the recoverable and dissipative powers are rewritten as:

− pTr ( d ) =
WR = − p1 : d =
σR : d
2WD = κ Tr 2 ( d ) + 2 µ d′ : d′ = σ D : d REMARK
For an incompressible fluid,
WR = − p Tr ( d ) =
0

26
Thermodynamic considerations
 Specific recoverable stress power is an exact differential,
1 1 dG
=
W σR =
:d → (exact differential)
ρ R
ρ dt

Then, the recoverable stress work per unit mass in a closed cycle is
zero: B ≡ A 1 B≡ A
1
B≡ A


A
ρ
WR dt = ∫
A
ρ
σ R : d dt = ∫
A
dG = GB ≡ A − GA = 0

 This justifies the denomination


“recoverable stress power”.

28
Thermodynamic Considerations
 According to the 2nd Law of Thermodynamics, the dissipative
power is necessarily non-negative,
2WD ≥ 0 2WD = κ Tr 2 ( d ) + 2 µ d′ : d′ = 0 d= 0

In a closed cycle, the work done by the dissipative stress per unit
mass will, in general, be different to zero:
B≡B
1

A
ρ
σ D : d dt > 0

2WD > 0

 This justifies the denomination “dissipative power”.

29
Limitations in the Viscosity Values
 The thermodynamic restriction,
2WD = κ Tr 2 ( d ) + 2 µ d′ : d′ ≥ 0
introduces limitations in the values of the viscosity parameters
κ , λ and µ :

1. For a purely spherical deformation rate tensor:


Tr ( d ) ≠ 0
2WD κ Tr 2 ( d ) ≥ 0
2
= κ =λ + µ ≥ 0
d′ = 0 3

2. For a purely deviatoric deformation rate tensor:


Tr ( d ) = 0
= µ d′ : d′ 2µ dij′ dij′ ≥ 0
2WD 2= µ≥0
d′ ≠ 0
>0

30
Chapter 9
Constitutive Equations in Fluids

rs
ee
s gin
9.1 Concept of Pressure

t d le En

r
Several concepts of pressure are used in continuum mechanics (hydrostatic pres-

ba
ge ro or
eS m
sure, mean pressure and thermodynamic pressure) which, in general, do not co-

ci
incide.
f

ra
C d P cs
b
a
i
9.1.1 Hydrostatic Pressure
an an n
y ha

le
liv or ec

Definition 9.1. Pascal’s law


M

.A

In a confined fluid at rest, the stress state on any plane containing a


m

given point is the same and is characterized by a compressive normal


d

stress.
uu
e
X Th

er
tin

In accordance with Pascal’s law, the stress state of a fluid at rest is characterized
on

.O

by a stress tensor of the type


C

σ = −p0 1
, (9.1)
σi j = −p0 δi j i, j ∈ {1, 2, 3}

where p0 is denoted as hydrostatic pressure (see Figure 9.1).

Definition 9.2. The hydrostatic pressure is the compressive normal


stress, constant on any plane, that acts on a fluid at rest.

439
440 C HAPTER 9. C ONSTITUTIVE E QUATIONS IN F LUIDS

Figure 9.1: Stress state of a fluid at rest.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
Figure 9.2: Mohr’s circle of the stress tensor of a fluid at rest.

ci
f

ra
C d P cs
b
a
i
an an n
y ha

Remark 9.1. The stress tensor of a fluid at rest is a spherical tensor


le
liv or ec

and its representation in the Mohr’s plane is a point (see Figure 9.2).
Consequently, any direction is a principal stress direction and the
M

.A

stress state is constituted by the state defined in Section 4.8 of Chap-


m

ter 4 as hydrostatic stress state.


d
uu
e
X Th

er
tin
on

.O

9.1.2 Mean Pressure


C

Definition 9.3. The mean stress σm is defined as


1 1
σm = σ ) = σii .
Tr (σ
3 3
The mean pressure p̄ is defined as minus the mean stress,
de f 1 1
σ ) = − σii .
p̄ = mean pressure = −σm = − Tr (σ
3 3

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Concept of Pressure 441

Remark 9.2. In a fluid at rest, the mean pressure p̄ coincides with the
hydrostatic pressure p0 ,
1
σ = −p0 1 =⇒ σm = (−3p0 ) = −p0 =⇒ p̄ = p0 .
3
Generally, in a fluid in motion the mean pressure and the hydrostatic
pressure do not coincide.

rs
ee
s gin
Remark 9.3. The trace of the Cauchy stress tensor is a stress invari-

t d le En
ant. Consequently, the mean stress and the mean pressure are also
stress invariants and, therefore, their values do not depend on the

r
ba
Cartesian coordinate system used.

ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n

9.1.3 Thermodynamic Pressure. Kinetic Equation of State


y ha

le
A new thermodynamic pressure variable, named thermodynamic pressure and
liv or ec

denoted as p, intervenes in the constitutive equations of fluids or gases.


M

.A
m

Definition 9.4. The thermodynamic pressure is the pressure variable


uu
e

that intervenes in the constitutive equations of fluids and gases, and


X Th

is related to the density ρ and the absolute temperature θ by means


er
tin

of the kinetic equation of state, F (p, ρ, θ ) = 0.


on

.O
C

Example 9.1
The ideal gas law is a typical example of kinetic equation of state:
F (p, ρ, θ ) ≡ p − ρRθ = 0 =⇒ p = ρRθ ,

where p is the thermodynamic pressure and R is the universal gas constant.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
442 C HAPTER 9. C ONSTITUTIVE E QUATIONS IN F LUIDS

Remark 9.4. In a fluid at rest, the hydrostatic pressure p0 , the mean


pressure p̄ and the thermodynamic pressure p coincide.

Fluid at rest : p0 = p̄ = p
Generally, in a fluid in motion the hydrostatic pressure, the mean
pressure and the thermodynamic pressure do not coincide.

rs
ee
s gin
Remark 9.5. A barotropic fluid is defined by a kinetic equation of
state in which the temperature does not intervene.

t d le En
Barotropic fluid : F (p, ρ) = 0 =⇒ p = f (ρ) =⇒ ρ = g (p)

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n

Remark 9.6. An incompressible fluid is a particular case of


y ha

barotropic fluid in which density is constant (ρ (x,t) = k = const.).


le
In this case, the kinetic equation of state can be written as
liv or ec
M

.A

F (p, ρ, θ ) ≡ ρ − k = 0
m

and does not depend on the pressure or the temperature.


d
uu
e
X Th

er
tin
on

.O

9.2 Constitutive Equations in Fluid Mechanics


C

Here, the set of equations, generically named constitutive equations, that must
be added to the balance equations to formulate a problem in fluid mechanics
(see Section 5.13 in Chapter 5) is considered. These equations can be grouped
as follows:
a) Thermo-mechanical constitutive equation
This equation expresses the Cauchy stress tensor in terms of the other ther-
modynamic variables, typically the thermodynamic pressure p, the strain
rate tensor d (which can be considered an implicit function of the velocity,
d (v) = ∇S v), the density ρ and the absolute temperature θ .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Constitutive Equation in Viscous Fluids 443

Thermo-mechanical
σ = −p 1 + f (d, ρ, θ ) 6 equations (9.2)
constitutive equation:

b) Entropy constitutive equation


An algebraic equation that provides the specific entropy s in terms of the
strain rate tensor, the density and the absolute temperature.

Entropy
s = s (d, ρ, θ ) 1 equation (9.3)

rs
constitutive equation:

ee
s gin
c) Thermodynamic constitutive equations or equations of state

t d le En
These are typically the caloric equation of state, which defines the specific
internal energy u, and the kinetic equation of state, which provides an equa-

r
ba
ge ro or
tion for the thermodynamic pressure.

eS m
ci
Caloric equation of f

ra
u = g (ρ, θ )
C d P cs
state:
b
a
Kinetic equation of 2 equations (9.4)
F (ρ, p, θ ) = 0
i
an an n

state:
y ha

le
liv or ec

d) Thermal constitutive equations


M

.A

The most common one is Fourier’s law, which defines the heat flux by con-
m

duction q as
d


uu

⎨ q = −k · ∇θ
e
X Th

Fourier’s
er

∂θ
tin

3 equations (9.5)
law: ⎩ qi = ki j i ∈ {1, 2, 3}
∂xj
on

.O

where k is the (symmetrical second-order) tensor of thermal conductivity,


C

which is a property of the fluid. For the isotropic case, the thermal conductiv-
©

ity tensor is a spherical tensor k = k 1 and depends on the scalar parameter k,


which is the thermal conductivity of the fluid.

9.3 Constitutive Equation in Viscous Fluids


The general form of the thermo-mechanical constitutive equation (see (9.2)) for
a viscous fluid is

σ = −p 1 + f (d, ρ, θ )
, (9.6)
σi j = −p δi j + fi j (d, ρ, θ ) i, j ∈ {1, 2, 3}

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
444 C HAPTER 9. C ONSTITUTIVE E QUATIONS IN F LUIDS

where f is a symmetrical tensor function. According to the character of the func-


tion f, the following models of fluids are defined:
a) Stokesian or Stokes fluid: the function f is a non-linear function of its argu-
ments.
b) Newtonian fluid: the function f is a linear function of its arguments.
c) Perfect fluid: the function f is null. In this case, the mechanical constitutive
equation is σ = −p1.
In the rest of this chapter, only the cases of Newtonian and perfect fluids will

rs
be considered.

ee
s gin
Remark 9.7. The perfect fluid hypothesis is frequently used in hy-
draulic engineering, where the fluid under consideration is water.

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
9.4 Constitutive Equation in Newtonian Fluids
i
an an n
y ha

The mechanical constitutive equation1 for a Newtonian fluid is


le
liv or ec

σ = −p 1 +C
C:d
M

.A

, (9.7)
σi j = −p δi j + Ci jkl dkl i, j ∈ {1, 2, 3}
m

d
uu

where C is a constant fourth-order (viscosity) constitutive tensor. A linear de-


e
X Th

pendency of the stress tensor σ on the strain rate tensor d is obtained as a result
er
tin

of (9.7). For an isotropic Newtonian fluid, the constitutive tensor C is an isotropic


on

.O

fourth-order tensor.

C

C = λ 1 ⊗ 1 + 2μI
©

  (9.8)
Ci jkl = λ δi j δkl + μ δik δ jl + δil δ jk i, j, k, l ∈ {1, 2, 3}

Replacing (9.8) in the mechanical constitutive equation (9.7) yields

σ = −p 1 + (λ 1 ⊗ 1 + 2μI) : d = −p 1 + λ Tr (d) 1 + 2μ d , (9.9)


which corresponds to the constitutive equation of an isotropic Newtonian fluid.

1 Note that the thermal dependencies of the constitutive equation are not considered here and,
thus, the name mechanical constitutive equations.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Constitutive Equation in Newtonian Fluids 445


Constit. eqn. of σ = −p 1 + λ Tr (d) 1 + 2μ d
an isotropic (9.10)
Newtonian fluid σi j = −p δi j + λ dll δi j + 2μ di j i, j ∈ {1, 2, 3}

Remark 9.8. Note the parallelism that can be established between the
constitutive equation of a Newtonian fluid and that of a linear elastic
solid (see Chapter 6):

rs
Newtonian fluid Linear elastic solid
 

ee
σ = −p 1 +CC:d σ =C:ε

s gin
σi j = −p δi j + Ci jkl dkl σi j = Ci jkl εkl

t d le En

r
ba
ge ro or
eS m
ci
f

ra
Remark 9.9. The parameters λ and μ physically correspond to the
C d P cs
b
a
viscosities, which are understood as material properties. In the most
i
an an n

general case, they may not be constant and can depend on other ther-
y ha

modynamic variables,
le
liv or ec

λ = λ (ρ, θ ) and μ = μ (ρ, θ ) .


M

.A

A typical example is the dependency of the viscosity on the temper-


m

ature in the form μ (θ ) = μ0 e−α(θ −θ0 ) , which establishes that the


d
uu
e

fluid’s viscosity decreases as temperature increases (see Figure 9.3).


X Th

er
tin
on

.O
C

Figure 9.3: Possible dependency of the viscosity μ on the absolute temperature θ .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
446 C HAPTER 9. C ONSTITUTIVE E QUATIONS IN F LUIDS

9.4.1 Relation between the Thermodynamic and Mean Pressures


In general, the thermodynamic pressure p and the mean pressure p̄ in a New-
tonian fluid in motion will be different but are related to each other. From the
(mechanical) constitutive equation of a Newtonian fluid (9.10),

σ = −p 1 + λ Tr (d) 1 + 2μ d =⇒

σ ) = −p Tr (1) + λ Tr (d) Tr (1) + 2μ Tr (d) = −3p + (3λ + 2μ) Tr (d) =⇒


Tr (σ


rs
−3 p̄ 2
p = p̄ + λ + μ Tr (d) = p̄ + K Tr (d)

ee
3


s gin
K
(9.11)

t d le En
where K is denoted as bulk viscosity.

r
ba
ge ro or
eS m
ci
2
f
Bulk viscosity : K = λ + μ

ra
(9.12)
C d P cs
3
b
a
i
an an n
y ha

Using the mass continuity equation (5.24), results in


le
liv or ec

dρ 1 dρ
+ ρ∇ · v = 0 =⇒ ∇·v = − (9.13)
ρ dt
M

.A

dt
m

Then, considering the relation


d
uu
e

∂ vi
X Th

Tr (d) = dii = = ∇·v (9.14)


er
tin

∂ xi
on

.O

and replacing in (9.11), yields


C

K dρ
p = p̄ + K∇ · v = p̄ − (9.15)
ρ dt

which relates the mean and thermodynamic pressures.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Constitutive Equation in Newtonian Fluids 447

Remark 9.10. In accordance with (9.15), the thermodynamic pres-


sure and the mean pressure in a Newtonian fluid will coincide in the
following cases:
• Fluid at rest: v=0
=⇒ p = p̄ = p0

• Incompressible fluid: = 0 =⇒ p = p̄
dt
• Fluid with null bulk viscosity K (Stokes’ condition2 ):

rs
2
K=0 =⇒ λ =− μ =⇒ p = p̄

ee
3

s gin
t d le En

r
ba
ge ro or
9.4.2 Constitutive Equation in Spherical and Deviatoric Components

eS m
ci
Spherical part f

ra
C d P cs
b
a
From (9.15), the following relation is deduced.
i
an an n
y ha

p̄ = p − K ∇ · v = p − K Tr (d) (9.16)
le
liv or ec
M

.A

Deviatoric part
m

Using the decomposition of the stress tensor σ and the strain rate tensor d
d
uu

in its spherical and deviator components, and replacing in the constitutive equa-
e
X Th

er

tion (9.10), results in


tin

1
on

.O

σ= σ ) 1 + σ  = − p̄ 1 + σ  = −p 1 + λ Tr (d) 1 + 2μ d =⇒
Tr (σ
3 

C

−3 p̄
 
©


σ = ( p̄ − p) 1 + λ Tr (d) 1 + 2μd = λ − K Tr (d) 1 + 2μd =⇒



−K Tr (d) 2
λ+ μ
3

 2 1
σ = − μ Tr (d) 1 + 2μd = 2μ d − Tr (d) 1 =⇒
3 3


d
(9.17)
2 Stokes’ condition is assumed in certain cases because the results it provides match the
experimental observations.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
448 C HAPTER 9. C ONSTITUTIVE E QUATIONS IN F LUIDS

σ  = 2μd (9.18)

where (9.16) and (9.12) have been taken into account.

9.4.3 Stress Power, Recoverable Power and Dissipative Power


Using again the decomposition of the stress and strain rate tensors in their spher-
ical and deviatoric components yields

rs
1
σ = − p̄ 1 + σ  d= Tr (d) 1 + d ,

ee
and (9.19)
3

s gin
and replacing in the expression of the stress power density (stress power per unit
of volume) σ : d, results in3

t d le En

r
1

ba
σ : d = (− p̄ 1 + σ  ) : 

ge ro or
Tr (d) 1 + d =

eS m
3

ci
f

ra
1 1
= − p̄ Tr (d) 1 : 1 +σ σ  : d − p̄ 1 : d + Tr (d) σ  : 1 =
C d P cs
b
a
3 

 3 

(9.20)
i
 
an an n

3 Tr d = 0 Tr σ = 0
y ha

= − p̄ Tr (d) + σ  : d .
le
liv or ec
M

.A

Replacing (9.16) and (9.17) in (9.20) produces


 
m

σ : d = − p − K Tr (d) Tr (d) + 2μ d : d .
d

(9.21)
uu
e
X Th

er
tin
on

.O

σ : d = −p Tr (d) + K Tr2 (d) + 2μ d : d = WR + 2WD


C



(9.22)
©

recoverable power dissipative power


WR 2WD

Recoverable power density: WR = −p Tr (d)


(9.23)
Dissipative power density: 2WD = K Tr2 (d) + 2μ d : d

3 The property that the trace of a deviator tensor is null is used here.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Constitutive Equation in Newtonian Fluids 449

Associated with the concepts of recoverable and dissipative powers, the re-
coverable and dissipative parts of the stress tensor, σ R and σ D , respectively, are
defined as

σ = − p 1 + λ Tr (d) 1 + 2μ d =⇒ σ = σR +σD . (9.24)





σR σD
Using the aforementioned notation, the recoverable, dissipative and total power
densities can be rewritten as

⎨ WR = −p Tr (d) = −p 1 : d = σ R : d ,

rs
ee
⎩ 2W = K Tr2 (d) + 2μ d : d = σ : d ,
D D (9.25)

s gin
σ R + σ D ) : d = σ R : d + σ D : d = WR + 2WD .
σ : d = (σ

t d le En

r
ba
ge ro or
eS m
ci
f
Remark 9.11. In an incompressible fluid, the recoverable power is

ra
C d P cs
null. In effect, since the fluid is incompressible, dρ/dt = 0 , and
b
a
i
considering the mass continuity equation (5.24),
an an n
y ha

1 dρ
∇·v = − = 0 = Tr (d) =⇒ WR = −p Tr (d) = 0 .
le
liv or ec

ρ dt
M

.A
m

d
uu
e
X Th

er
tin

Remark 9.12. Introducing the decomposition of the stress


power (9.25), the balance of mechanical energy (5.73) becomes
on

.O

  
dK dK
C

Pe = + σ : d dV = + σ R : d dV + σ D : d dV
©

dt dt
V V V

 
dK
Pe = + WR dV + 2WD dV ,
dt
V V

which indicates that the mechanical power entering the fluid Pe is


invested in modifying the kinetic energy K and creating recoverable
power and dissipative power.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
450 C HAPTER 9. C ONSTITUTIVE E QUATIONS IN F LUIDS

9.4.4 Thermodynamic Considerations


1) It can be proven that, under general conditions, the specific recoverable
power (recoverable power per unit of mass) is an exact differential
1 1 dG
WR = σ R : d = . (9.26)
ρ ρ dt
In this case, the recoverable work per unit of mass performed in a closed cycle
will be null (see Figure 9.4),

rs

B≡A 
B≡A 
B≡A
1 1
WR dt = σ R : d dt = dG = GB≡A − GA = 0 ,

ee
(9.27)
ρ ρ

s gin
A A A

which justifies the denomination of WR as recoverable power.

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

Figure 9.4: Closed cycle.


uu
e
X Th

er
tin

2) The second law of thermodynamics allows proving that the dissipative power
on

.O

2WD in (9.25) is always non-negative,


C

2WD ≥ 0 ; 2WD = 0 ⇐⇒ d=0 (9.28)


©

and, therefore, in a closed cycle the work performed per unit of mass by the
dissipative stresses will, in general, not be null,
B
1
σ D : d dt > 0 . (9.29)
ρ 

A
2WD > 0
This justifies the denomination of 2WD as (non-recoverable) dissipative power.
The dissipative power is responsible for the dissipation (or loss of energy) phe-
nomenon in fluids.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Constitutive Equation in Newtonian Fluids 451

Example 9.2 – Explain why an incompressible Newtonian fluid in motion


that is not provided with external power (work per unit of time) tends to
reduce its velocity to a complete stop.

Solution
The recoverable power in an incompressible fluid is null (see Remark 9.11).
In addition, the dissipative power 2WD is known to be always non-negative
(see (9.28)). Finally, applying the balance of mechanical energy (see Re-
mark 9.12) results in

rs
 
dK
0 = Pe = + WR dV + 2WD dV =⇒

ee
dt 

s gin
V =0 V

 
dK

t d le En
d 1 2
= 2 WD dV < 0
ρv dV = −
dt dt 2 

r
ba
>0

ge ro or
V V

eS m
ci
f
and, therefore, the fluid looses (dissipates) kinetic energy and the velocity of

ra
C d P cs
its particles decreases.
b
a
i
an an n
y ha

le
9.4.5 Limitations in the Viscosity Values
liv or ec
M

.A

Due to thermodynamic considerations, the dissipative power 2WD in (9.25) has


been seen to always be non-negative,
m

d
uu

2WD = K Tr2 (d) + 2μ d : d ≥ 0 . (9.30)


e
X Th

er
tin

This thermodynamic restriction introduces limitations in the admissible values


on

.O

of the viscosity parameters K, λ and μ of the fluid. In effect, given a certain


fluid, the aforementioned restriction must be verified for all motions (that is,
C

for all velocity fields v) that the fluid may possibly have. Therefore, it must be
©

verified for any arbitrary value of the strain rate tensor d = ∇S (v). Consider, in
particular, the following cases:
a) The strain rate tensor d is a spherical tensor.
In this case, from (9.30) results
Tr (d) = 0 ; d = 0 =⇒ 2WD = K Tr2 (d) ≥ 0 =⇒
2 (9.31)
K=λ+ μ ≥0
3
such that only the non-negative values of the bulk viscosity K are feasible.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
452 C HAPTER 9. C ONSTITUTIVE E QUATIONS IN F LUIDS

b) The strain rate tensor d is a deviatoric tensor.


This type of flow is schematically represented in Figure 9.5. In this case,
from (9.30) results
Tr (d) = 0 ; d = 0 =⇒ 2WD = 2μ d : d = 2μ d  i j : d  i j ≥ 0 =⇒


>0 (9.32)
μ ≥0

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
⎡ ⎤
i
⎡ ⎤ 1 ∂ vx
an an n

0 0
y ha

vx (y) ⎢ 2 ∂y ⎥
⎢ ⎥ ⎢ ⎥
v (x, y) = ⎢ ⎥ ; ⎢ ∂ ⎥
le
d=⎢ 1 v 
0 ⎥=d
x
liv or ec

⎣ 0 ⎦ 0
⎣ 2 ∂y ⎦
M

.A

0
0 0 0
m

d
uu
e

Figure 9.5: Flow characterized by a deviatoric strain rate tensor.


X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
CH.10. FLUID MECHANICS
Multimedia Course on Continuum Mechanics
Overview
 Governing Equations
 Newtonian Fluids Lecture 1
 Barotropic fluids

 Hydrostatics. Fluids at rest


Lecture 2
 Hydrostatic problem
 Archimedes´ Principle Lecture 3
 Equilibrium of Floating Solids Lecture 4

 Barotropic Perfect Fluids


Lecture 5
 Fluid Mechanics Equations
 Bernoulli’s Trinomial Lecture 6
 Steady State Solution Lecture 7
 Transient State Solution Lecture 8
Lecture 9
2
Overview (cont’d)
 Newtonian Viscous Fluids
 Navier-Stokes Equations
 Energy Equation Lecture 10
 Reduced System of Equations
 Physical Interpretations
 Reduced System of Equations for Particular Cases Lecture 11

 Boundary Conditions
 BC in velocities
 BC in pressures Lecture 12
 Mixed BC
 BC on free surfaces

 Laminar and Turbulent Flows Lecture 13

3
10.1. Governing Equations
Ch.10. Fluid Mechanics

4
Reminder – Governing Eqns.
 Balance equations of the thermo-mechanical problem:
Conservation of Mass.
ρ + ρ ∇ ⋅ v = 0 Continuity Equation.
1 eqn.

Linear Momentum Balance.


∇ ⋅ σ + ρ b = ρ v 3 eqns.
Cauchy’s Motion Equation.
Angular Momentum Balance. Symmetry
8 PDE + σ =σT of Cauchy’s Stress Tensor.
3 eqns.
2 restrictions
ρ u σ : d + ρ r − ∇ ⋅ q
= Energy Balance.
1 eqn.
First Law of Thermodynamics.

− ρ ( u − θ s ) + σ : d ≥ 0 Clausius-Planck
Inequality. Second Law of
1 2 restrictions
− 2 q ⋅ ∇θ ≥ 0 Heat flux Thermodynamics.
ρθ Inequality.

5
Newtonian Fluids
 Constitutive equations of the thermo-mechanical problem in a
Newtonian fluid:
Thermo-Mechanical
− p1 + λTr ( d ) 1 + 2µ d
σ= Constitutive Equations.
6 eqns.

Entropy
s = s ( d, ρ , θ ) Constitutive Equation. 1 eqn.
12 PDE
Thermal Constitutive Equation. Fourier’s
q = − K θ Law of Conduction. 3 eqns.

u = f ( ρ ,θ ) Caloric
State Equations. 2 eqns.
F ( ρ , p, θ ) = 0 Kinetic

 Grand total of 20 PDE with 20 unknowns:


ρ → 1, v → 3, σ → 9, u → 1, q → 3, θ → 1, s → 1, p → 1

6
Barotropic Fluids
 A barotropic fluid is characterized by the
kinetic state equation:
F ( ρ , p, θ ) = 0 ρ = ρ ( p)
 The uncoupled mechanical problem becomes:
Conservation of Mass.
ρ + ρ ∇ ⋅ v = 0 Continuity Equation.
1 eqn.

Linear Momentum Balance.


∇ ⋅ σ + ρ b = ρ v 3 eqns.
First Cauchy’s Motion Equation.

Thermo-Mechanical
− p1 + λTr ( d ) 1 + 2µ d
σ= Constitutive Equations.
6 eqns.

ρ = ρ ( p) Kinetic State Equation 1 eqn.

 11 scalar unknowns: ρ , v , σ , p . (Considering the symmetry of Cauchy


stress tensor, σ , will have 6 unknowns).
7
10.2. Hydrostatics. Fluids at Rest
Ch.10. Fluid Mechanics

8
Hydrostatic stress state vs. Hydrostatic
problem
 Uniform velocity, v ( x ,t ) ≡ v ( t ) :⇒ v =  ⊗ v = v ⊗  = 0
1
d = ∇S v = [v ⊗ ∇ + ∇ ⊗ v ] = 0
2
HYDROSTATIC
Thus, − p1 + λTr ( d ) 1 + 2 µ d
σ= σ=− p 1 ⇒ p =p STRESS SATE
=0 =0
 Uniform and stationary velocity, v ( x ,t ) ≡ cnt :
dv ∂v
a= = + v ⋅∇ v= 0
dt ∂t
HYDROSTATIC
Thus, σ = − p0 1 Tr ( σ ) = −3 p0 p= p= p0 CASE

 Fluid at rest, v ( x ,t ) ≡ cnt =


0 . A particular hydrostatic case (where the name
comes from)

9
Hydrostatic Problem
 A hydrostatic problem ( v ( x ,t ) ≡ cnt ) is characterized by:
Conservation of Mass.
ρ + ρ ∇ ⋅ v = 0 ρ ( X, t ) = ρ0 ( X ) 1 eqn.
Continuity Equation.

Linear Momentum Balance.


 ⋅σ + ρ b =ρ v  ⋅σ + ρ b =
0 3 eqns.
First Cauchy’s Motion Equation.

Thermo-Mechanical
− p1 + λTr ( d ) 1 + 2µ d
σ= σ = − p0 1 Constitutive Equations.
6 eqns.

 Substituting the constitutive and the continuity eqn. into the Cauchy
eqn.:
−∇p0 + ρ0b =0 FUNDAMENTAL

σ =− p0 1 ⇒ ∇ ⋅ σ =∇ ⋅ (− p0 1) =−∇p0 →  ∂p0 EQUATION OF
− ∂x + ρ0=
bi 0 i ∈ {1, 2,3} HYDROSTATICS
 i

10
Gravity forces. Triangular pressure
distribution
 For a fluid subjected to gravity forces,
−∇p0 + ρ0b =0 0
  
 ∂p0 b ( x, t ) =  0 
− ∂x + ρ0=
bi 0 i ∈ {1, 2,3} − g 
 i  
the momentum eq. can be written as:
∂p0 ( x, y , z ) ∫ dx
− =
0 p0 ( x, y, z ) ≡ p0 ( y, z )
∂x
∂p0 ( y , z ) ∫ dy p0 ( y, z ) ≡ p0 ( z )
− =
0
∂y
dp0 ( z ) ∫ dz − ρ0 gz + C
p0 =
− − ρ0 g =
0
dz
 If the surface pressure is considered zero, then: =p0 ρ0 g ( h − z )

=
p0 z = h 0 − ρ0 g=
h+C 0 =C ρ0 g h Triangular
pressure
distribution
11
Archimedes´ Principle
 Any fluid applies a buoyant force (up-thrust) to an object that is
partially or completely immersed in it.
 The magnitude of the buoyant force is equal to
the weight of the fluid displaced by the object.
 The resultant of the buoyant force on a floating
object acts at the center of mass of the displaced
fluid (center of buoyancy).

12
Archimedes´ Principle - Proof
 Consider a solid with a volume V and density ρ within a fluid in a
hydrostatic case. Then,
 The traction vector on the solid boundary :
t =σ ⋅ n =− p0 1 ⋅ n =− p0 n
p0 ( z ) ρ0 g ( h − z )
=

 The resultant force exerted by the fluid


on the solid :

R= ∫ t dS= ∫ − p ( z )n dS
∂V ∂V
0

R only depends on the hydrostatic pressure


distribution on the boundary of the solid

13
Archimedes´ Principle R= ∫ t dS= ∫ − p ( z )n dS
0

(first part proof) ∂V

−∇p0 + ρ 0b =
0
∂V

 Consider the same fluid without the solid in it, and replaced by fluid .
Then,
 Pressures on the boundary of the “replacing” fluid are the same than in the
immersed solid case (and, therefore, the resulting force, R)
Volume of the displaced
 The divergence theorem can be applied: fluid=V
(The pressure distribution is now continuous in space)
 0 
0   0 
 
R = ∫ − p0 ndS = ∫ −∇p0 dV ≡ − ρ 0  0  ∫ dV =  
 ρ
∂V V −ρ b − g  
V  gV 

0
0
V  W 
b
 Finally, R= E eˆ z= W eˆ z ⇒ E= W
Up-thrust on the body ( E ) = weight of the
fluid displaced by the body (W )
14
Archimedes´ Principle −∇p0 + ρ 0b =
0
(second part proof)
 Consider the moment of the up-thrust forces at the center of mass
(center of gravity, CG) of the volume of displaced fluid:
Divergence
Theorem
M GE = ∫ x × ( − p n ) dS= ∫ x × (− 
0 p  ) dV =
0
∂V V
=∇p0

− ∫ ( x × p0 ) dV
=
V

 Substituting the fundamental eq. of hydrostatics ( ∇p0 = ρ0b ),

− ∫ ( x × ρ0b ) dV =
M GE = −MWG =
0
V
= MWG Moment of the weight of the displaced fluid with
respect to its center of gravity (by definition it must
be zero)
The up-thrust force, E, passes through the CG
of the volume of the displaced fluid (center of buoyancy).
15
Equilibrium of Floating Solids
 The equilibrium can be:

 Stable: the solid’s CG


is below the center of
buoyancy (CG of the
displaced fluid).

 Unstable: the solid’s CG


is above the center of
buoyancy (CG of the
displaced fluid).

16
10.3. Fluid Dynamics.
Barotropic Perfect Fluids
Ch.10. Fluid Mechanics

17
Barotropic Perfect Fluids
 A perfect fluid is a Newtonian fluid with null viscosity, µ= λ= 0 :

− p1 + λTr ( d ) 1 + 2µ d
σ= σ = − p1 hydrostatic
stress state

 Therefore,  ⋅ σ =−p
σ:d = − pTr ( d )
− p1 : d =

 In a barotropic fluid temperature does not intervene in the kinetic


state equation:
F ( ρ , p, θ ) = 0 ρ = ρ ( p)
REMARK
Do not confuse a hydrostatic stress
state (spherical stress tensor) with
a hydrostatic flow regime (null or
uniform velocity).

18
Barotropic Perfect Fluids: Field Equations
 The mechanical problem for a barotropic perfect fluid:

Conservation of Mass.
ρ + ρ  ⋅ v =0
Continuity Equation.
1 eqn.

Linear Momentum Balance.


 ⋅σ + ρ b =ρ v −p + ρ b =
ρ v Euler’s Equation.
3 eqns.

F ( ρ , p, θ ) = 0 ρ = ρ ( p) Kinetic State Equation 1 eqn.

 5 scalar unknowns: ρ , v , p .

19
Barotropic Perfect Fluids: Field
Equations
 The thermal problem for a barotropic perfect fluid:

Thermal Constitutive Equation. Fourier’s 3 eqns.


q = q (θ ) = − K θ Law of Conduction.

ρ u= σ : d + ρ r −  ⋅ q Energy Balance. 1 eqn.


ρ u =− p ⋅ v + ρ r + K  θ2 First Law of Thermodynamics.

u = f ( ρ , v, θ , p ) u = u ( ρ ,θ ) Caloric State Equation 1 eqn.

 Once the mechanical problem is solved, the thermal problem can


be calculated as there are 5 scalar unknowns: u → 1, q → 3, θ → 1

20
Bernoulli’s Trinomial
 Consider a barotropic fluid with potential body forces:
0
T
 ∂φ ∂φ ∂φ  0
φ(
x, t ) =
gz → b ( x, t ) =
−φ ( x, t ) =
− =
  ∂x ∂y ∂z   
Body forces  − g 
potential

 And, consider the following lemmas:


 Lemma 1. For a barotropic fluid there exists a function P (x, t ) = P
ˆ ( p (x, t ))
which satisfies:
p 1 1
p = ρ P Proof : P ( p ) ≡ ∫ dp P = p
0 ρ ( p ) ρ ( p)
 Lemma 2. The convective term of the acceleration can be written as:
1 
v ⋅ ∇v = 2ω × v + ∇  v 2 
2 
Where 2ω
=  × v is the vorticity vector.
21
Bernoulli’s Trinomial
 Taking the Euler equation and substituting,

1 dv
−p + ρ =
b ρ v → − p + =
b
ρ 
dt
∂v + v⋅∇v
∂v
∂t
−P − φ= + 2ω × v +  ( 12 v 2 )
1 ∂t
 p = P ; b = − ∇φ
ρ
v ⋅ v = 2ω × v +  12 v 2 )

Bernoulli’s Trinomial
 Rearranging, ∂v EQ. OF MOTION for a
−   P + φ + 12 v 2 = + 2ω × v barotropic perfect fluid
  ∂t under potential body forces

22
Barotropic perfect fluid with potential forces:
Steady state solution
 The equation of motion for a steady flow becomes:
∂v  1 
−   P + φ + 12 v 2 = + 2ω × v − ∇  P + φ + v 2=
 2ω × v
  ∂t  2 

 Considering a stream line Γ : x =


x( s ) parameterized in terms of
its arc length s :

23
Barotropic perfect fluid with potential forces:
Steady state solution
 Then, the equation of motion along the considered streamline , Γ , reads:
dx ds
=
= M ( x( s ) ) ds dt
−   P + φ + 12 v 2  = 2ω × v
  − ( M ) ⋅ v= ( 2ω × v ) ⋅ v
=0

∇M ( x ( s ) ) ⋅ = = 0
dx dM
∀x ∈ Γ
 ds ds M=
(x) cnt ∀x ∈ Γ
∂M
∂x

Γ: x =
x( s )

( )
Beurnoulli’s trinomial  P + φ + 12 v 2 =x cnt ∀x ∈ Γ
remains constant along 
the same streamline.

24
Barotropic perfect fluid with potential
(gravitational)forces: steady state solution
 Incompressible fluid:
ρ ( p) =
p 1 1 p p
ρ= ρ0 =
cnt P( p) ≡ ∫ dp = ∫0 
dp =
0 ρ
 ( )
p ρ 0 ρ0
ρ0
 Potential gravitational forces:
0
b (x) = 0
−φ ( x ) = φ (x) =
gz
 
 − g 
 p 
 P + φ + 12 v 2  (=
 )  ρ
x + gz + 2 v=
1 2
 cnt ∀x ∈ Γ
 
 0

p 1 v 2 def BERNOULLI’S
z+ + = H= cnt ∀x ∈ Γ THEOREM
ρ0 g 2 g

25
Barotropic perfect fluid with potential
(gravitational)forces: steady state solution
 Bernoulli’s Theorem can be interpreted as:
pressure velocity
head head
elevation p 1 v 2 def
z+ + = H= cnt ∀x ∈ Γ
ρ0 g 2 g
velocity total or
head energetic
Piezometric or head
pressure hydraulic head, h
head
elevation

 It is a statement of the conservation of mechanical energy:


p 1 2
+ gz + = v cnt ∀x ∈ Γ
pressure ρ 0 2
energy potential kinetic
energy energy
26
Example
Determine the velocity and mass flow rate of water from the circular
hole (0.1m diameter) at the bottom of the water tank (at this instant).
The tank is open to the atmosphere and h = 4m. Consider a steady
state regime.

27
Example - Solution
p1 v12 p2 v22
z1 + + =z2 + +
S1 = cross section area at 1 ρ0 g 2 g ρ0 g 2 g
 p=
1 p=
2 patm ≈ 0

 S2
h  v1 ≈ 0 ( S1 ≥ S2 → v=
1 v 2 ≈ 0)
 S
1

 ≈0

h
v22
S 2 = cross section area at 2 z=
1 z2 + =v2 2g ( z1 − z2 )
2g
Velocity at the bottom hole of the tank: v 2 = 2gh

28
Barotropic perfect fluid with potential forces:
Transient solution
 The equation of motion for an unsteady flow is:

∂v
−   P + φ + 12 v 2 = + 2ω × v
  ∂t

 This expression may be simplified for:


 Potential (irrotational) flow
 Potential and incompressible flow REMARK
A movement is said to be irrotational
(or potential) if the rotational of the
velocity field is null at any point:

∇ × v( x,=
t) 0 ∀x , ∀t

29
Barotropic perfect fluid with potential forces
and irrotational flow: Transient solution
 In an irrotational flow:
1
 × v( x, t )= 0 ∀x , ∀t ω( x, t )=
 × v( x, t )= 0 ∀x , ∀t
2
 There will exist a scalar function (named velocity potential χ ( x, t ) ) which
satisfies: v ( x,t ) = χ ( x, t )

 Then, the equation of motion becomes:


∂v ∂χ
−   P + φ + 12 v 2 = + 2ω × v −   P + φ + 12 v 2  = )
  ∂t   ∂t
∂ ( ∇χ ( x, t ) )
=
 Rearranging, ∂t
= M ( x, t )
 ∂χ 
−   P + φ + 12 v 2 +  = 0 ∇M ( x, t )= 0 ∀x , ∀t
 ∂t 
30
Barotropic perfect fluid with potential forces
and irrotational flow: Transient solution
 The momentum equation can be trivially integrated:
∂χ
M ( x, t )= 0 ∀x , ∀ t M ( x, t ) = P + φ + 12 v 2 + = ϕ (t )
∂t

 Defining a modified velocity potential, χ ( x, t ) :

t ∇= χ v ( x,t )
χ ∇=
( x, t ) χ ( x, t ) − ∫ ϕ (τ ) dτ
def
χ= ∂χ ∂χ
0 = − ϕ (t )
∂t ∂t
 Finally,
∂χ Differential equation of
P + φ + ( ∇χ ) + = 0
1 2
∀x , ∀t
2 ∂t hydraulic transients

31
Barotropic perfect fluid with potential forces:
Transient solution in irrotational flows
 The mechanical problem for a potential (irrotational) flow:
v ( x,t ) = ∇χ ( x, t )

Conservation of Mass.
ρ + ρ  ⋅ (χ ( x, t ) ) =
0 ρ + ρ 2 χ =
0 1 eqn.
Continuity Equation.

∂χ
P (ρ, p) + φ + ( ) ∂t = 0
1 2 Linear Momentum Balance.
 χ + ∀x , ∀ t 1 eqn.
2 Hydraulic Transients Equation.

F ( ρ , p, θ ) = 0 ρ = ρ ( p) Kinetic State Equation 1 eqn.


barotropic fluid

 3 scalar unknowns: ρ, p , χ . Once the potential χ is obtained,


the velocity field can be easily calculated:
v ( x,t ) = χ ( x, t )

32
Incompressible perfect fluid with potential
forces: transient solution in irrotational flows
 In an incompressible flow:

=0 ρ = ρ0
dt

 Then, the term P in the equation of motion becomes:


p p

( x, t ) ∫ ρ =
P=
1 1
dp = ∫ dp
p
0 ( p) ρ 0 0 ρ0

 And, the equation of motion is:


∂χ
p 1
( )
2
−  +φ +  χ + ]=0
ρ0 2 ∂t

33
Fluid Mechanics Equations
 The mechanical problem for a potential (irrotational) and
incompressible flow:

Conservation of Mass.
ρ + ρ  ⋅ (χ ( x, t ) ) =
not
0 ∇ 2 χ =∆χ =0 Continuity Equation.
1 eqn.

∂χ
p 1
( ) Linear Momentum Balance.
2
+φ +  χ + = 0 ∀x , ∀ t 1 eqn.
ρ0 2 ∂t Hydraulic Transients Equation.

 2 scalar unknowns: p , χ . Once the potential χ is obtained, the


velocity field can be easily calculated:
v ( x,t ) = χ ( x, t )

34
10.4. Newtonian Viscous Fluids
Ch.10. Fluid Mechanics

35
Governing Equations
 Governing equations of the general fluid mechanics problem:
Conservation of Mass.
ρ + ρ  ⋅ v =0 Continuity Equation. 1 eqn.

Linear Momentum Balance.


 ⋅σ + ρ b =ρ v Equation of Motion. 3 eqns.

Energy Balance.
ρ u= σ : d + ρ r −  ⋅ q First Law of Thermodynamics. 1 eqn.

Mechanical Constitutive
− p1 + λTr ( d ) 1 + 2 µ d
σ= Equations.
6 eqns.

Entropy
s = s ( d, θ , ρ ) Constitutive Equation.
1 eqn.

q = − K θ Thermal Constitutive Equation. Fourier’s


Law of Conduction. 3 eqns.

F ( ρ ,θ , p ) = 0 Caloric and Kinetic State


u = u ( ρ ,θ ) Equations. 2 eqns.

Too large to solve


 17 scalar unknowns: ρ, v , σ , u , q , θ , s , p . analytically !!
36
Navier-Stokes Equations  ⋅σ + ρ b =ρ v

 Consider the following lemmas:


 Lemma 1.
 ⋅ d = ∆v +  ( ⋅ v )
1 1 Where d( x, t ) is the
2 2 deformation rate tensor

 Lemma 2.
 ⋅ (α 1 ) =
α Where α ( x, t ) is a scalar function.

 Introducing the constitutive equation into the divergence of the


stress tensor, ⋅ σ ,and taking into account these lemmas:

− p1 + λ Tr ( d ) 1 + 2 µ d
σ=  ⋅ σ=  ⋅ ( − p1 + λTr (d )1 + 2 µ d )=

= ⋅v =−p + λ (Tr (d )) + µ∆v + µ( ⋅ v )
 
(⋅v )
37
Navier-Stokes Equations
 Then, the linear momentum balance equation is rearranged:

−p + ( λ + µ )  ( ⋅ v ) + µ∆v + ρ b =
dv dv
 ⋅σ + ρb =ρ ρ
dt dt


 − p + ( λ + µ )  (  ⋅ v ) + µ∆ v + ρ b = ρ
dv
 dt
 NAVIER-STOKES
∂ ∂2v j ∂ 2
+ (λ + µ )
− p vi dv
+µ = + ρ bi ρ i i, j ∈ {1, 2,3} EQUATIONS
 ∂xi ∂xi ∂x j ∂x j ∂x j dt

 The Navier-Stokes equations are essentially the equation of motion
(Cauchy’s equation) expressed solely in terms of velocity and pressure.

38
Energy Equation du
ρ = σ : d + ρ r −⋅q
dt

 Consider the definition of stress power:

− pTr ( d ) + κ Tr 2 ( d ) + 2 µ d′ : d′
σ :d =
RECOVERABLE DISSIPATIVE
POWER, WR . POWER, 2WD .

 And the Fourier’s Law:


q = − K θ  ⋅ q =− ⋅ ( K θ )

 Introducing these into the energy balance equation:


ρ u= σ : d + ρ r −  ⋅ q
du
ρ ( pTr (d ) + κ Tr 2 (d ) + 2 µd′ : d′) + ρ r +  ⋅ ( K θ )
=−
dt
39
Energy Equation
 Then, the energy balance equation is reduced to:
ρ
du
dt
=−
( pTr
 ( d ) + κ Tr 2
( d ) + 2 µd′ : d′) + ρ r +  ⋅ ( K θ )
⋅v
DISSIPATIVE
POWER, 2WD.
 du
 ρ dt =− p ⋅ v + ρ r +  ⋅ ( K θ ) + κ Tr ( d ) + 2 µ d′ : d′
2


 du ∂v ∂ ∂θ ∂v
ρ =− p i + ρr + (K ) + κ ( i ) 2 + 2 µ dij′ dij′ i, j ∈ {1, 2,3}
 dt ∂xi ∂xi ∂xi ∂xi

ENERGY EQUATION
 The energy equation is essentially the energy balance equation expressed
solely in terms of velocity and pressure.

40
Reduced System of Equations
 Governing equations of the general fluid mechanics problem are
reduced to:
Conservation of Mass.
ρ + ρ  ⋅ v =0 Continuity Equation. 1 eqn.
Momentum Balance.
−p + ( λ + µ )  ( ⋅ v ) + µ∆v + ρ b =
ρ v Navier-Stokes Equations. 3 eqns.

ρ u =− p ⋅ v + ρ r +  ⋅ ( K θ ) + κ Tr 2 ( d ) + 2 µ d′ : d′ Energy Balance. 1 eqn.

Entropy
s = s ( v, θ , ρ ) Constitutive Equation.
1 eqn.

u = u ( ρ ,θ ) Caloric State Equation. 1 eqn.

F ( ρ ,θ , p ) = 0 Kinetic State Equation. 1 eqn.

 8 scalar unknowns: ρ , v , p , u , θ, s .REMARK


For a barotropic fluid, the mechanic and
thermal problems are uncoupled, reducing
41 the mechanical problem to 5 unknowns.
Navier-Stokes Equations: Physical
Interpretation
 The Navier-Stokes equations can be physically interpreted as:


{  
}
−p − − ( λ + µ )  ( ⋅ v ) + µ∆v + ρ b − ρ
dv
dt
=0
>0
Forces due to Viscous forces due
the pressure Body Inertial
to the contact with forces forces
gradient neighbour particles

NOTE: All forces are


per unit of volume.

42
Energy Equation: Physical
Interpretation
 The energy equation can be physically interpreted as:
du
ρ =− p ⋅ v + ρ r +  ⋅ ( K θ ) + κ Tr 2 (d ) + 2 µ d′ : d′
dt

Variation of Mechanical work of the Dissipative


internal energy thermodynamic pressure power, 2WD.
per unit of time:
d  Heat generated by the
( dV ) = (∇ ⋅ v )dV 
dt  1 d internal sources and
 − p∇ ⋅ v =− p ( dV ) conduction per unit
1 d dt dV
∇⋅v = ( dV )  volume per unit time
dV dt 
Variation of volume
per unit of volume
and per unit of NOTE: All terms are
time per unit of volume
and unit of time

43
Fluid Mechanics Equations ∂ρ
+  ⋅ ( ρ v) =
0
in Curvilinear Coordinates ∂t

 CONTINUITY EQUATION
 Cartesian Coordinates
∂ρ ∂ ∂ ∂
+ ( ρ vx ) + ( ρ v y ) + ( ρ vz ) =
0
∂t ∂x ∂y ∂z

 Cylindrical Coordinates
∂ρ 1 ∂ 1 ∂ ∂
+ ( r)
ρ r v + ( θ)
ρ v + ( ρ vz ) =
0
∂t r ∂r r ∂θ ∂z

 Spherical Coordinates
∂ρ 1 ∂ ∂ ∂
+ 2 ( ρ r 2 vr ) + ( ρ vϕ ) =
1 1
( ρ vθ sin θ ) + 0
∂t r ∂r r sin θ ∂θ r sin θ ∂ϕ

44
Fluid Mechanics Equations =aρa = − ∇p + ( λ + µ ) ( )
∇ ∇ ⋅ v + µ∆v + ρ b
∂v
+ v ⋅ ∇v
in Curvilinear Coordinates
; ∇⋅v =0
∂t

 NAVIER-STOKES EQUATIONS for an incompressible fluid


with ρ and µ constants
 Cartesian Coordinates
 x component
 ∂v x ∂v x ∂v x ∂v x  ∂p  ∂2vx ∂2vx ∂2vx 
ρ + vx + vy + vz =− +µ 2 + 2 + 2  + ρ bx
 ∂t ∂x ∂y ∂z  ∂x  ∂x ∂y ∂z 
 y component

 ∂v y ∂v y ∂v y ∂v y  ∂p  ∂2v y ∂2v y ∂2v y 


ρ + vx + vy + vz =− +µ 2 +
 ∂x
+ 2  + ρ by
 ∂t ∂x ∂y ∂z  ∂y  ∂ y 2
∂z 
 z component
 ∂v z ∂v z ∂v z ∂v z  ∂p  ∂2vz ∂2vz ∂2vz 
ρ + vx + vy + vz =− +µ 2 + 2 + 2  + ρ bz
 ∂t ∂x ∂y ∂z  ∂z  ∂x ∂y ∂z 

NOTE: For “slow” motions left-hand-side term is zero ( ρ a = 0)


45
Fluid Mechanics Equations =aρa = − ∇p + ( λ + µ ) ( )
∇ ∇ ⋅ v + µ∆v + ρ b
∂v
+ v ⋅ ∇v
in Curvilinear Coordinates
; ∇⋅v =0
∂t

 NAVIER-STOKES EQUATIONS for an incompressible fluid


with ρ and µ constants
 Cylindrical Coordinates
 r component
 ∂v r ∂v r vθ ∂v r vθ2 ∂v r  ∂p  ∂ 1 ∂  1 ∂ v r 2 ∂vθ ∂ v r
2 2

ρ + vr + − + vz =− +µ  ( rv r )  + 2 2 − 2 + 2  + ρ br
 ∂t ∂r r ∂θ r ∂z  ∂r ∂
 r r ∂r  r ∂θ r ∂θ ∂z 
 θ component
 ∂v ∂v v ∂v vv ∂v  1 ∂p  ∂ 1 ∂  1 ∂ vθ 2 ∂v r ∂ vθ
2 2

ρ  θ + vr θ + θ θ + r θ + v z θ =− +µ  ( rvθ )  + 2 2 + 2 + 2  + ρ bθ
 ∂t ∂r r ∂θ r ∂z  r ∂θ  ∂r  r ∂r  r ∂θ r ∂θ ∂z 
 z component
 ∂v ∂v v ∂v ∂v  ∂p  1 ∂  ∂v z  1 ∂ vz ∂ vz
2 2

ρ  z + vr z + θ z + v z z  =− +µ r + 2 + 2  + ρ bz
 ∂t ∂r r ∂θ ∂z  ∂z  r ∂ r  ∂r  r ∂θ 2
∂z 

NOTE: For “slow” motions left-hand-side term is zero ( ρ a = 0)


46
Fluid Mechanics Equations =aρa = − ∇p + ( λ + µ ) ( )
∇ ∇ ⋅ v + µ∆v + ρ b
∂v
+ v ⋅ ∇v
in Curvilinear Coordinates
; ∇⋅v =0
∂t

 NAVIER-STOKES EQUATIONS for an incompressible fluid


with ρ and µ constants
 Spherical Coordinates
 ∂v r ∂v r vθ ∂v r vϕ ∂ v r vθ2 + vϕ2  ∂p
 r component ρ  + vr + + −  =
− +
 ∂t ∂r r ∂θ r sin θ ∂ϕ r  ∂r
 ∂ 1 ∂  ∂  ∂v r  ∂ 2 vr ∂ 2 ∂vϕ 
+ µ   2 ( r 2 vr )  + 2
1 1 2
 sin θ + 2 2 − 2 ( vθ sin θ ) − 2  + ρ br
 ∂r  r ∂r  r sin θ ∂θ  ∂θ  r sin θ ∂ϕ r sin θ ∂θ r sin θ ∂ϕ 
2

 θ component  ∂vθ
ρ  + vr
∂vθ vθ ∂vθ
+ +
vϕ ∂vθ v r vθ vϕ2 cotg θ
+ −
 1 ∂p
 = − +
 ∂t ∂r r ∂θ r sin θ ∂ϕ r r  r ∂θ
 1 ∂  2 ∂vθ  1 ∂  1 ∂  1 ∂ 2 vθ 2 ∂v r 2 cotg θ ∂vϕ 
+µ  2  r + 2  ( vθ sin θ )  + 2 2 + 2 − 2  + ρ bθ
 r ∂r  ∂r  r ∂θ  sin θ ∂θ  r sin θ ∂ ϕ 2
r ∂θ r sin θ ∂ϕ 
 φ component ρ  ∂vϕ + v ∂vϕ + vθ ∂vϕ + vϕ ∂vϕ + vϕ v r + vθ vϕ cotg θ  =

1 ∂p
+
 
 ∂t ∂r r ∂θ r sin θ ∂ϕ r sin θ ∂ϕ
r
r r 
 1 ∂  2 ∂vϕ  1 ∂  1 ∂  ∂ 2 vϕ ∂v r 2 cotg θ ∂vθ 
+µ  2  r
 r ∂r ∂
+
 2
∂θ 
 θ ∂θ
( vϕ sin θ ) 
+
 2 2
1
θ ∂ ϕ 2
+ 2
2
θ ∂ϕ
+ 2
r sin θ ∂ϕ
 + ρ bϕ
  r  r sin r sin r sin 
47
Fluid Mechanics Equations σ=− p1 + λTr ( d ) 1 + 2 µ d

in Curvilinear Coordinates
2
λ= K - µ ; Tr (d= ) ⋅ v
3

 STRESS TENSOR for Newtonian fluids 2


(K= λ + µ )
 Cartesian Coordinates 3

 ∂v 2   ∂v x ∂v y 
σ x µ  2 x − (·v )  − p + K ·v
= τ xy = τ yx = µ + 
 ∂x 3   ∂y ∂x 

 ∂v y 2   ∂v y ∂v z 
σ y = µ 2 − (∇·v ) − p + K ∇·v τ yz = τ zy = µ + 
 ∂y 3   ∂z ∂y 

 ∂v 2   ∂v ∂v 
σ z = µ 2 z − (∇·v ) − p + K ∇·v τ zx = τ xz = µ  z + x 
 ∂z 3   ∂x ∂z 

NOTE: for incompressible fluids ( ⋅ v =0)

48
Fluid Mechanics Equations σ=− p1 + λTr ( d ) 1 + 2 µ d

in Curvilinear Coordinates
2
λ= K - µ ; Tr (d= ) ⋅ v
3

STRESS TENSOR for Newtonian fluids 2


 (K= λ + µ )
 Cylindrical Coordinates 3

 ∂v 2 
σ r µ  2 r − (·v )  − p + K ·v
=  ∂  v  1 ∂v r 
τ rθ = τ θr = µ r  θ  +
 ∂r 3  
 ∂r  r  r ∂θ 

  1 ∂vθ v r  2   ∂v 1 ∂v z 
=σ θ µ 2  +  − ( ·v )  − p + K ·v τ θz = τ zθ = µ  θ +
  r ∂θ r  3   ∂z r ∂θ 

 ∂v 2   ∂v ∂v 
σ z µ  2 z − (·v )  − p + K ·v
= τ zr = τ rz = µ  z + r 
 ∂z 3   ∂r ∂z 

1 ∂ 1 ∂vθ ∂v z
=
·v ( rv r ) + +
r ∂r r ∂θ ∂z

49
NOTE: for incompressible fluids ( ⋅ v =0)
Fluid Mechanics Equations σ=− p1 + λTr ( d ) 1 + 2 µ d

in Curvilinear Coordinates
2
λ= K - µ ; Tr (d= ) ⋅ v
3

STRESS TENSOR for Newtonian fluids 2


 (K= λ + µ )
 Spherical Coordinates 3
 ∂v 2   ∂  v  1 ∂v r 
σ r µ  2 r − (·v )  − p + K ·v
= τ rθ = τ θr = µ r  θ  + 
 ∂r 3   ∂r  r  r ∂θ 

  1 ∂vθ v r  2   sinθ ∂  v φ  1 ∂v θ 
=σ θ µ 2  +  − ( ·v )  − p + K ·v τ θφ = τ φθ = µ  
 sinθ  + r sinθ ∂φ 
  r ∂θ r  3   r ∂θ   

  1 ∂vϕ v r vθ cotg θ  2 
σϕ µ 2  + + −
 3 ( ·v )  − p + K ·v
  r sin θ ∂ϕ r r    1 ∂v r ∂  vφ 
τ φr = τ rφ = µ  + r  

 r sinθ ∂φ ∂r  r 
1 ∂ 2 ∂ 1 ∂ vϕ
·v =
r 2 ∂r
( r v r ) + r sin θ ∂θ ( vθ sin θ ) + r sin θ ∂ϕ
1

50
NOTE: for incompressible fluids ( ⋅ v =0)
10.5. Boundary Conditions
Ch.10. Fluid Mechanics

61
Boundary Conditions in Velocities
 Prescribed velocities
 Velocities are known in a certain part of the control volume
boundary, Γ v :=v ( x, t ) v ( x, t ) ∀x ∈ Γ
v

 Impervious walls
 Part of the boundary of control volume, Γ vn , which can be mobile, is
impervious (it cannot be penetrated by the fluid).
 The normal component of the relative fluid/wall velocity, v r ≡ v − v* , is
considered null.

( v − v )=
*
⋅n 0 ∀x ∈ Γ v n

v n ( x,t ) = v ⋅ n = v* ⋅ n ∀x ∈ Γ v n

63
Boundary Conditions in Velocities
 Adherent walls
 In a viscous fluid in contact with a wall the fluid is considered to adhere to
the wall.
 The relative fluid/wall velocity, v r , is considered null.

v r ( x, t ) = v − v* = 0 ∀x ∈ Γ v

=
v v* ∀x ∈ Γ v

64
Boundary Conditions in Pressures
 Prescribed tractions
 The traction vector’s value is prescribed in certain parts of the control
volume contour Γσ :

t ( x, t )= σ ⋅ n= t* ( x, t ) ∀x ∈ Γσ

 Sometimes, only part of the traction


vector is prescribed, such as the
thermodynamic pressure.
 For a Newtonian fluid:
− p1 + λTr (d )1 + 2 µd
σ= t =σ ⋅ n =− pn + λTr (d )n + 2 µd ⋅ n

=
p ( x , t ) p * ( x , t ) ∀x ∈ Γ p

65
Mixed Boundary Conditions
 Prescribed traction vector and velocities
 Pressure and the tangential component of the velocity, v t , are prescribed :

 This boundary condition is typically used in problems involving in-flow and


out-flow sections (pipes).

66
Boundary Conditions on Free Surfaces

 The contact surface between air and fluid (generally water) is a


free surface.

67
Boundary Conditions on Free Surfaces

 HYPOTHESIS: The free surface is a material surface.


 This implicitly establishes certain boundary conditions on the velocity field
of the material surface Γ fs.
 Consider the free surface:
=Γ fs : {x | φ ( x, y , z, t ) ≡=
z − η ( x, y , t ) 0}
 Impose the condition for a material surface (null material derivative):
d φ ∂φ ∂η ∂η ∂η ∂φ
= + v ⋅ φ =− − vx − vy + vz =0
dt ∂t ∂t ∂x ∂y ∂z = 1

∂η ∂η ∂η
v z ( x,=
t) + vx + vy ∀x ∈ Γ fs
∂t ∂x ∂y

68
Boundary Conditions on Free Surfaces

 Another boundary condition typically used on free surfaces is:

p (=
x, t ) Patm ∀x ∈ Γ fs

 This allows identifying the position of the free surface once the
pressure field is known:

=Γ fs : {x | p ( x,=
t ) − Patm 0 }

69
10.6. Laminar and Turbulent Flows
Ch.10. Fluid Mechanics

70
Laminar Flow
 Flow persists as unidirectional movement.
 Particles flow in parallel layers which do not mix.
 A flow’s laminar character is identified by the Reynolds number: Re < 1000
 The governing equations of the fluid mechanics problem are valid for this
type of flow.

def
V ×L
Re =
ν

V , Flow’s characteristic velocity


L , Domain’s characteristic length
ν , Kinematic viscosity:

71
Turbulent Flow
 High values of the Reynolds number.
 Highly distorted and unstable flow:
 Stress and velocity at a given spatial point fluctuate randomly and very
fast, along time, about a mean value.
 Specific models (turbulence models) are used to characterize this regime.

NOTE: Turbulent flow is


out of the scope of this
course

72
Chapter 10
Fluid Mechanics

rs
ee
s gin
10.1 Governing Equations

t d le En

r
A fluid is a particular case of continuous medium that is characterized by its

ba
ge ro or
eS m
specific set of constitutive equations. Consequently, the fluid mechanics problem

ci
is defined by the following equations:
f

ra
C d P cs
b
a
a) Balance Equations
i
an an n

1) Mass continuity equation


y ha

le
liv or ec


+ ρ∇ · v = 0 (1 equation) (10.1)
M

dt
.A
m

2) Balance of linear momentum


uu
e

dv
X Th

er

∇·σ +ρ b = ρ
tin

(3 equations) (10.2)
dt
on

.O

3) Energy balance
C

du
ρ = σ : d+ρ r−∇·q (1 equation) (10.3)
dt

4) Restrictions imposed by the second law of thermodynamics


 
Clausius-Planck du ds
−ρ −θ +σ : d ≥ 0
inequality dt dt (10.4)
Heat conduction 1
− 2 q · ∇θ ≥ 0
inequality ρθ

453
454 C HAPTER 10. F LUID M ECHANICS

b) Constitutive Equations
5) Thermo-mechanical constitutive equation

σ = −p 1 + λ Tr (d) 1 + 2μ d (6 equations) (10.5)

6) Entropy constitutive equation

s = s (d, ρ, θ ) (1 equation) (10.6)

rs
ee
7) Law of heat conduction

s gin
q = −K ∇θ (1 equation) (10.7)

t d le En

r
c) Thermodynamic equations of state

ba
ge ro or
eS m
ci
8) Caloric equation of state f

ra
C d P cs
b
a
u = u (ρ, θ )
i
(1 equation) (10.8)
an an n
y ha

le
liv or ec

9) Kinetic equation of state


M

.A

F (ρ, p, θ ) = 0 (1 equation) (10.9)


m

The unknowns1 of these governing equations are


uu
e


X Th

er

ρ → 1 unknown ⎪
tin



v → 3 unknowns ⎪ ⎪


on

.O

σ → 6 unknowns ⎪ ⎪



C

u → 1 unknown
→ 17 unknowns .
©

(10.10)
q → 3 unknowns ⎪ ⎪

θ → 1 unknown ⎪ ⎪



s → 1 unknown ⎪ ⎪


p → 1 unknown

The system is formed by a total of 17 PDEs and 17 unknowns which, in general,


should be solved together, that is, in a coupled form. However, as noted in Sec-
tion 5.13.1 of Chapter 5, under certain hypotheses or situations a reduced system

1 Note that the strain rate tensor d is not considered an unknown since it is an implicit function
of the velocity field v.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Governing Equations 455

of equations, denoted as the mechanical problem, may be posed and solved sep-
arately for a reduced number of unknowns (mechanical variables).
Consider the case of a barotropic fluid, which is characterized by the fact that
the temperature does not intervene in the kinetic equation of state. Then,
Kinetic equation F (ρ, p) = 0 =⇒ ρ = ρ (p) , (10.11)
of state

which establishes that the density


may be described solely by means

rs
of the thermodynamic pressure (see

ee
Figure 10.1). Assuming, in addition,
that the temperature does not inter-

s gin
vene in the thermo-mechanical consti-
tutive equation (10.5), the governing

t d le En
equations of the (uncoupled) mechani-
Figure 10.1: Density depends on

r
cal problem in a Newtonian fluid are de-

ba
ge ro or
the thermodynamic pressure in a

eS m
fined as

ci
barotropic fluid.
f

ra
C d P cs
b
a
1) Mass continuity equation
i
an an n


y ha

+ ρ∇ · v = 0 (1 equation) (10.12)
le
dt
liv or ec
M

.A

2) Cauchy’s equation
m

dv
∇·σ +ρ b = ρ
uu

(3 equations) (10.13)
e

dt
X Th

er
tin

3) Mechanical constitutive equation


on

.O
C

σ = −p 1 + λ Tr (d) 1 + 2μ d (6 equations) (10.14)


©

4) Kinetic equation of state

ρ = ρ (p) (1 equation) (10.15)


The unknowns of the problem posed by the equations above are

ρ → 1 unknown ⎪


v → 3 unknowns
→ 11 unknowns . (10.16)
σ → 6 unknowns ⎪ ⎪

p → 1 unknown

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
456 C HAPTER 10. F LUID M ECHANICS

A reduced system of 11 equations and 11 unknowns (mechanical problem) is


obtained, which may be solved uncoupled from the rest of the problem (thermal
problem).

10.2 Hydrostatics. Fluids at Rest


Consider the following particular cases in terms of a fluid’s velocity:
a) Uniform velocity: v (x,t) ≡ v (t)
In this case, the spatial description of the velocity does not depend on the

rs
spatial point being considered and is only a function of time. Therefore,

ee
1

s gin
d = ∇S v =(v ⊗ ∇ + ∇ ⊗ v) = 0 . (10.17)
2

t d le En
Then, the constitutive equation (10.14) is reduced to

r
σ = −p 1 + λ Tr (d) 1 + 2 μ d =⇒ σ = −p1 ,

ba
ge ro or
(10.18)

eS m



ci
=0 =0 f

ra
C d P cs
b
a
which indicates that the stress state is hydrostatic (see Figure 10.2). In addi-
i
an an n

tion, the mean pressure p̄ and the thermodynamic pressure p coincide,


y ha

σ ) = −3 p̄ = −3p
Tr (σ =⇒ p̄ = p .
le
(10.19)
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 10.2: Mohr’s circle for a fluid with uniform velocity.

b) Uniform and stationary velocity: v (x,t) ≡ const.


A fluid with uniform and stationary velocity is characterized, in addition of
(10.17), by

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Hydrostatics. Fluids at Rest 457


dv ∂ v ⎪

a= = + v · ∇v = 0 hydrostatic
dt ∂t (10.20)
⎪ case
σ = −p0 1 =⇒ p̄ = p = p0 ⎭

This is the most general case of hydrostatics, which is characterized by a null


acceleration (the velocity of each particle is constant, although not necessar-
ily null) and the three pressures (thermodynamic p, mean p̄, and hydrostatic
p0 ) coincide.

rs
ee
c) Fluid at rest: v (x,t) ≡ const. = 0

s gin
A particular case of hydrostatics is that of a fluid at rest with null velocity.

t d le En

r
10.2.1 Hydrostatic Equations

ba
ge ro or
eS m
ci
f
The hydrostatic problem is governed by the following equations:

ra
C d P cs
b
a
1) Constitutive equation
i
an an n
y ha

σ = −p0 1
(10.21)
le
σi j = −p0 δi j i, j ∈ {1, 2, 3}
liv or ec
M

.A

where p0 is the hydrostatic pressure.


m

d
uu
e

Remark 10.1. Pascal’s Principle states that, in a fluid at rest, the


X Th

er
tin

pressure is the same in every direction.


This classic fluid mechanics postulate is guaranteed by the spheri-
on

.O

cal structure of the stress tensor in (10.21), which ensures that all
C

directions are principal stress directions (see Figure 10.3).


©

Figure 10.3: Pascal’s Principle.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
458 C HAPTER 10. F LUID M ECHANICS

2) Mass continuity equation



dρ ⎪

+ ρ∇ · v = 0 dρ
dt ⇒ =0 ⇒ ρ (X,t) = ρ0 (X) = const.
⎪ dt
v = const. ⇒ ∇ · v = 0 ⎭
(10.22)
and the density of a same particle does not change along time.

3) Cauchy’s equation

rs
ee
dv
∇·σ +ρ b = ρ (10.23)

s gin
dt
Introducing (10.21) and (10.22) in (10.23),

t d le En



r
⎨ ∇ · σ = ∇ · (−p0 1) = −∇p0

ba
ge ro or
eS m
ci
⎪    
⎪ ∂ σi j ∂ ∂ p0 f

ra

⎩ ∇·σ j = = (−p0 δi j ) = − = − ∇p0 j j ∈ {1, 2, 3}
C d P cs
∂ xi ∂ xi ∂xj
b
a
i
an an n

(10.24)
y ha

le
liv or ec



⎨ −∇p0 + ρ0 b = 0
M

.A

Fundamental equation (10.25)


of hydrostatics ⎩ − ∂ p0 + ρ0 bi = 0

m

i ∈ {1, 2, 3}
d

∂ xi
uu
e
X Th

er
tin
on

.O

10.2.2 Gravitational Force. Triangular Pressure Distribution


C

Consider the particular case, which is in fact very common, of the body forces
©

b (x,t) corresponding to the gravitational force (assumed constant in space and


along time, and oriented in the negative direction of the x3 -axis, as shown in
Figure 10.4).
Since the acceleration is null (see (10.20)) it is a quasi-static problem and, be-
cause the actions b (x,t) ≡ const. are independent of time, so are the responses,
in particular, the hydrostatic pressure. Consequently,
p0 (x,t) ≡ p0 (x) = p0 (x, y, z) , (10.26)
and (10.25) can be integrated as follows

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Hydrostatics. Fluids at Rest 459

rs
Figure 10.4: Gravitational Force.

ee
s gin


⎪ ∂ p0 (x, y, z)

⎪ − =0 =⇒ p0 (x, y, z) ≡ p0 (y, z)

t d le En

⎪ ∂x

r
∂ p0 (y, z)

ba
ge ro or
− =0 =⇒ p0 (y, z) ≡ p0 (z) (10.27)

eS m

⎪ ∂y

ci


⎪ f

ra

⎩ − ∂ p0 (z) − ρ0 g = 0

C d P cs
=⇒ p0 = −ρ0 g z +C
b
a
∂z
i
an an n
y ha

For a case such as the one shown in Figure 10.5, in which the surface pressure
le
(height z = h) is considered null, the solution (10.26) results in
liv or ec


M

.A


p0  = 0 ⇒ −ρ0 g h +C = 0 ⇒ C = ρ0 g h ⇒ p0 = ρ0 g (h − z) ,
z=h
m

(10.28)
uu
e

which corresponds to a triangular pressure distribution, as shown in Figure 10.5.


X Th

er
tin
on

.O
C

Figure 10.5: Pressure distribution on a gravity dam.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
460 C HAPTER 10. F LUID M ECHANICS

10.2.3 Archimedes’ Principle

Definition 10.1. Archimedes’ principle:


1) The upward buoyant force experienced by a body submerged in
a fluid is equal to the weight of the fluid displaced by said body.
The classical principle is complemented with:
2) The resultant of the aforementioned buoyant force acts at the

rs
center of gravity of the volume of the displaced fluid.

ee
s gin
To prove Archimedes’ principle, consider the situations in Figure 10.6. On the

t d le En
one hand, Figure 10.6 a) illustrates a solid with volume V and density ρ in
the interior of a fluid of density ρ0 . The solid is not necessarily in equilibrium,

r
ba
ge ro or
even though its velocity and acceleration are assumed to be small enough to

eS m
ci
ensure a hydrostatic state in the fluid. On the other hand, Figure 10.6 b) shows
f

ra
the same fluid without the solid, such that the volume occupied by said solid in
C d P cs
b
a
Figure 10.6 a) is occupied here by an identical volume of fluid.
i
an an n
y ha

le
liv or ec

Volume of
M

.A

displaced fluid
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 10.6: a) Solid submerged in a fluid and b) volume of the displaced fluid.

a) Pressure and stress distributions in the fluid


Using the fundamental equation of hydrostatics (10.25) and considering that
the gravitational forces act in the negative direction of the z-axis, the situation
corresponding with (10.26) and (10.27) is achieved. Thus, the result (10.28) is
valid for both cases a) and b) of Figure 10.6.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Hydrostatics. Fluids at Rest 461

p0 (z) = ρ0 g (h − z)
(10.29)
σ = −p0 1

Note that the hydrostatic pressure and the stress state in the fluid are the same
for equivalent points of the fluid in the cases a) and b) of Figure 10.6.

b) Buoyant force on the submerged solid


The traction vector on the boundary of the submerged solid in Figure 10.6 a) is

rs
t = σ · n = −p0 1 · n = −p0 n (10.30)

ee
s gin
and the resultant R of the forces the fluid exerts on the solid is
 

t d le En
R= t dS = −p0 n dS . (10.31)

r
∂V ∂V

ba
ge ro or
eS m
ci
Note now that, since the hydrostatic pressure distribution is the same in both
f

ra
cases of Figure 10.6, this resultant is the same as the one obtained in case b)
C d P cs
b
a
for the forces that the rest of the fluid exerts on the volume of displaced fluid,
i
an an n

with the particularity that, because the pressure distribution is constant in space
y ha

(with value p0 ), the Divergence Theorem (Stokes’ Theorem) can be applied


le
on (10.30), resulting in
liv or ec

 
M

.A

R= −p0 n dS = −∇p0 dV . (10.32)


m

∂V V
d
uu
e

Introducing (10.25) in (10.32) yields


X Th

er
tin

  
R= −∇p0 dV = −p0 b dV = − ρ0 b dV = W êz = E êz , (10.33)
on

.O

V V V


C

W êz
where E is the upward buoyant force acting on the submerged solid and W is the
weight of the displaced fluid (see Figure 10.6 b) ). That is,
upward buoyant force = weight of the displaced fluid , (10.34)



E W
whereby the first part of Archimedes’ Principle is proven.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
462 C HAPTER 10. F LUID M ECHANICS

Volume of
displaced fluid

rs
ee
s gin
Figure 10.7: Forces acting on the volume of displaced fluid.

t d le En

r
ba
ge ro or
eS m
c) Vertical line of application of the upward buoyant force

ci
f

ra
Consider now the moment MG
C d P cs
E of the upward buoyant force E with respect to
b
a
the center of gravity, G, of the volume of displaced fluid (see Figure 10.72 ),
i
an an n


y ha


⎪ 
Divergence
⎪ Theorem 
le

liv or ec


⎪ G = × (−p = × (−p ∇) = − x × ∇p0 dV
⎪ M
⎪ E x 0 n) dS x 0 dV

M

.A




⎪ ∂V V V

⎪  
⎨  G ∂  
m

ME i = − ei jk x j p0 nk dS = − ei jk x j p0 dV =
d
uu

∂ xk
e


⎪ ∂V
⎪ V
X Th


er

⎪   
tin


⎪ ∂xj ∂ p0 ∂ p0

⎪ = − e p dV − e x dV = − ei jk x j dV


i jk
∂ xk
0 i jk j
∂ xk ∂ xk
on

.O


⎪ V 



V V
⎩ ei jk δ jk = i ∈ {1, 2, 3}
C

ei j j = 0
©

(10.35)
and replacing the fundamental equation of hydrostatics (10.25) in (10.35) finally
yields
 
MG
E =− (x × ∇p0 ) dV = − (x × ρ0 b) dV = −MW
G
=0, (10.36)
V V


G
MW

2 Without loss of generality, the origin of the system of Cartesian axes can be placed at G.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Hydrostatics. Fluids at Rest 463

where MW G is the moment of the weight of the displaced fluid with respect to its
center of gravity G, which, considering the definition of the center of gravity,
is null. Consequently, the moment of the upward buoyant force E with respect
to the center of gravity of the volume of displaced fluid is also null. Then, it
is concluded that the vertical line of application of the upward buoyant force
crosses said center of gravity, as established by the second part of Archimedes’
principle.

Example 10.1 – Apply Archimedes’ principle to the study of stability of the


equilibrium in floating solids to determine how the relative position of the

rs
centers of gravity of the solid and the corresponding volume of displaced

ee
fluid affect the nature of this equilibrium.

s gin
Solution

t d le En
Consider a floating medium, in equilibrium, and the following two situations:

r
ba
a) The center of gravity of the solid (center of thrust) is below the center of

ge ro or
eS m
gravity of the displaced fluid (center of buoyancy).

ci
f

ra
C d P cs
In this case, any perturbation (inclination) tends to create a moment M = W d
b
a
i
in the sense that tends to recover the initial state of equilibrium. It is, thus, a
an an n

case of stable flotation equilibrium.


y ha

center of gravity
le
recovering
liv or ec

of the displaced
fluid moment
M

.A

perturbation
m

d
uu
e
X Th

er
tin

center of gravity
on

.O

of the solid
C

b) The center of gravity of the solid (center of thrust) is above the center of
gravity of the displaced fluid (center of buoyancy).
In this case, any perturbation (inclination) tends to create a moment M = W d
in the sense that tends to capsize the floating solid, that is, it tends to move
the solid further away from the initial state of equilibrium. It is, thus, a case
of unstable flotation equilibrium.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
464 C HAPTER 10. F LUID M ECHANICS

center of gravity capsizing


of the solid moment
perturbation

center of gravity
of the displaced fluid

rs
ee
Placing weights (ballasts) on the keel of a boat responds to the search of

s gin
improved flotation stability of this boat.

t d le En

r
ba
ge ro or
eS m
10.3 Fluid Dynamics: Barotropic Perfect Fluids

ci
f

ra
In the most common case, the velocity is not uniform nor stationary (v ≡ v (x,t)),
C d P cs
b
a
and, therefore, in general, the acceleration will not be null (a (x,t) = 0). In con-
i
an an n

sequence,
 the divergence  of the velocity (∇ · v = 0) and the gradient of the ve-
y ha

not
locity ∇ ⊗ v = ∇v = 0 will not be null either.
le
liv or ec
M

.A

Definition 10.2. A perfect fluid is a Newtonian fluid characterized


m

by the fact that the viscosities λ and μ (see (10.14)) are null.
d
uu
e
X Th

er
tin

The mechanical constitutive equation (10.14) of a perfect fluid becomes


on

.O


C

σ = −p 1 + λ Tr (d) + 2μ d
=⇒ σ = −p 1
©

λ =μ =0
(10.37)

∇ · σ = −∇ p
=⇒
σ : d = −p 1 : d = −p Tr (d)

which results in a hydrostatic stress state3 .

3 A hydrostatic stress state (the stress tensor is spherical) should not be confused with a
hydrostatic motion regime (the velocity is uniform or null).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Fluid Dynamics: Barotropic Perfect Fluids 465

Definition 10.3. A barotropic fluid is characterized by a kinetic


equation of state (10.9) in which the temperature does not intervene.

F (ρ, p, θ ) ≡ F (ρ, p) = 0 =⇒ ρ = ρ (p)

10.3.1 Equations of the Problems

rs
Taking into account the hypotheses of a perfect and a barotropic fluid, the equa-

ee
tions governing a fluid dynamics problem are reduced to:

s gin
a) Mechanical problem

t d le En
1) Mass continuity equation

r
ba
ge ro or
eS m

ci
+ ρ∇ · v = 0 (1 equation) (10.38)
f

ra
dt
C d P cs
b
a
i
an an n

2) Balance of linear momentum (Euler’s equation)


y ha

le
dv
liv or ec

−∇p + ρb = ρ (3 equations) (10.39)


dt
M

.A
m

3) Kinetic equation of state


d
uu
e
X Th

ρ = ρ (p) (1 equation) (10.40)


er
tin


The mechanical
 problem is composed of 5 equations and 5 unknowns ρ (x,t) , v (x,t),
on

.O

p (x,t) that can be solved uncoupled from the thermal problem.


C

b) Thermal problem
1) Fourier’s law

q = −K ∇θ ⇒ ∇ · q = −K ∇ · (∇θ ) = −K ∇2 θ (3 equations)
(10.41)
2) Energy balance
du
ρ = −p∇ · v +ρ r + K∇2 θ (1 equation)
dt 

(10.42)
σ :d −∇ · q

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
466 C HAPTER 10. F LUID M ECHANICS

3) Caloric equation of state

u = u (ρ, θ ) (1 equation) (10.43)



The thermal
 problem is defined by 5 equations and 5 unknowns q (x,t), θ (x,t),
u (x,t) and can be solved once the mechanical problem has been solved and the
velocity field v (x,t), the density ρ (x,t) and the pressure p (x,t) are known.

Remark 10.2. A general format of the fluid mechanics problem in-

rs
cludes the thermal conductivity K between the viscosities (in a gen-

ee
eralized sense) of the problem. The definition of a perfect fluid as
a fluid without viscosity results, in this context, in the cancella-

s gin
tion of the thermal conductivity (K = 0), therefore (10.41) leads to
q = −K∇θ = 0 and the thermal problem is reduced to the equa-

t d le En
tions (10.42) and (10.43).

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
10.3.2 Resolution of the Mechanical Problem under Potential Body
an an n

Forces. Bernoulli’s Trinomial


y ha

le
Consider now the mechanical problem for the particular case of potential body
liv or ec

forces (the body forces derive from a potential φ ),


M

.A
m

Potential body forces: b (x,t) = −∇φ (x,t) . (10.44)


d
uu
e
X Th

In the particular case of a gravitational potential with the line of action along the
er
tin

negative direction of the z-axis, the potential is


on

.O

⎡ ⎤
C

0
not ⎢ ⎥
©

φ (x, y, z,t) = g z =⇒ b = −∇φ ≡ ⎣ 0 ⎦ . (10.45)


−g

Lemma 10.1. For a barotropic fluid (ρ = ρ (p)) there exists a func-


tion P (x,t) = P̂ (p (x,t)) that satisfies
∇p = ρ ∇P .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Fluid Dynamics: Barotropic Perfect Fluids 467

Proof
Defining the function P (x,t) as
p
1
P (x,t) = P̂ (p (x,t)) = d p̄ , (10.46)
ρ ( p̄)
0
then, it will satisfy


⎪ ∂ P (x,t) ∂ P̂ ∂ p
⎨ =
∂ xi ∂ p ∂ xi (10.47)

⎪   ∂ P̂   1  

rs
⎩ ∇P i = ∇p i = ∇p i i ∈ {1, 2, 3}
∂p ρ (p)

ee
s gin
leading to
1
∇P = ∇p .

t d le En
(10.48)
ρ (p)

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
Lemma 10.2. The convective term of the acceleration can be written
i
an an n

as  
y ha

1 2
ω ×v+∇
v · ∇v = 2ω v ,
le
2
liv or ec

ω = ∇ × v is the vorticity vector.


M

.A

where 2ω
m

d
uu
e

Proof
X Th

er

Expanding the right-hand term in the Lemma4 ,


tin

 
  ∂vj ∂ v j ∂ vi ∂ vi ∂ vi
on

.O

v · ∇v j = vi = vi − +vi = 2 vi w ji +vi =
∂ xi ∂ xi ∂ x j ∂xj 
∂xj
C


−wi j
©

2 w ji
∂ vi ∂ vi
= −2 vi wi j + vi = 2 ei jk vi ωk +vi =
∂xj 
∂xj (10.49)
 e jkivi ω k   
∂ 1   1 2
= 2 e jki vi ωk + vi vi = 2ω ω ×v j + ∇ v ,

∂ x j 2 
2 j
ω × v] j
[2ω v · v = v2
j ∈ {1, 2, 3}
4The following results,
 previously obtained in Chapter 2, are used here:
w ji = −wi j = ∇a v ji = (∂ v j /∂ xi − ∂ vi /∂ x j ) /2 , wi j = −ei jk ωk and v2 = |v|2 = v · v.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
468 C HAPTER 10. F LUID M ECHANICS

which leads to  
1 2
ω ×v+∇
v · ∇v = 2ω v . (10.50)
2

Rearranging now Euler’s equation (10.39),


dv 1 dv
− ∇p + ρb = ρ =⇒ − ∇p + b = , (10.51)
dt ρ dt
and replacing (10.45) and (10.48) in (10.51) produces

rs
 

ee
dv ∂ v ∂v 1 2
− ∇P − ∇φ = = + v · ∇v = ω ×v+∇
+ 2ω v , (10.52)

s gin
dt ∂t ∂t 2

t d le En
where the result (10.50) has been taken into account. Finally, (10.52) is rewritten
as   

r
∂v

ba
1 2

ge ro or
− ∇P + ∇φ + ∇ = + 2ωω ×v .

eS m
v (10.53)
∂t

ci
2
f

ra
C d P cs
b
a
i
an an n

Equation of motion of a barotropic perfect fluid


y ha

under potential body forces


le
 
liv or ec

1 2 ∂v (10.54)
−∇ P + φ + v = + 2ωω ×v
M

.A

2 ∂t


m

Bernoulli’s trinomial
d
uu
e
X Th

er
tin

Equation (10.54) is the particular form adopted by the balance of linear momen-
tum (Euler’s equation (10.39)) in barotropic perfect fluids subjected to potential
on

.O

body forces.
C

10.3.3 Solution in a Steady-State Regime


The solution to the mechanical problem defined by (10.38) to (10.40) has, in
general, a transient regime, in which the spatial description of the mechanical
variables evolves along time, and a steady-state regime, in which said spatial
description is, approximately, constant along time (see Figure 10.8).
Consider now the equation of motion (10.54) in a steady-state regime,
 
∂v 1
=0 =⇒ −∇ P + φ + v2 = 2ω ω ×v , (10.55)
∂t 2

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Fluid Dynamics: Barotropic Perfect Fluids 469

Transient regime Steady-state regime

Figure 10.8: Transient and steady-state regimes.

rs
ee
and a streamline5 Γ : x = x (s) parametrized in terms of its arc-length s (see

s gin
Figure 10.9). Projecting (multiplying) equation (10.53) in the direction tangent
to the streamline, t, results in

t d le En
 
1 2

r
− ∇ P + φ + v = 2ω ω × v =⇒ − (∇M) · v = (2ω ω × v) · v = 0 (10.56)

ba
ge ro or
eS m
2 


ci

=0
dx ds f

ra
M (x)
C d P cs
b
a
ds dt
i

an an n


y ha

dx dM ⎪

∇M (x (s)) · = = 0⎪

le
liv or ec

ds ds ∀x ∈ Γ =⇒ M (x) = const. ∀x ∈ Γ

M


.A

∂ M (x (s)) dxi dM ⎪
= =0⎪ ⎭
∂ xi ds ds
m

(10.57)
uu
e

and (10.57) is written as


X Th

er
tin


1
on

.O

P + φ + v2 (x) = const. ∀x ∈ Γ , (10.58)


2
C

which establishes that Bernoulli’s trinomial remains constant along a same


streamline Γ .

Remark 10.3. Note that (10.58) is no longer a partial differential


equation but a (scalar) algebraic equation, already integrated. This
equation allows, thus, determining one of the unknowns of the me-
chanical problem once the others are known.

5 In a steady-state (stationary) regime, trajectories and streamlines coincide.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
470 C HAPTER 10. F LUID M ECHANICS

rs
Figure 10.9: Parametrized streamline.

ee
s gin
10.3.3.1 Solution in Steady-State Regime for an Incompressible Fluid under

t d le En
Gravitational Forces
Consider now the particular case of a barotropic fluid with the following char-

r
ba
ge ro or
acteristics:

eS m
ci
f

ra
a) The fluid is incompressible
C d P cs
b
a
i
ρ = ρ (p) = ρ0 = const. (10.59)
an an n
y ha

In this case, the function P (p) in (10.46) can be integrated as follows.


le
liv or ec

p p
1 1 p
M

.A

P (x,t) = d p̄ = d p̄ = (10.60)
ρ ( p̄) ρ0 ρ0
m

0 0
d
uu
e
X Th

b) The body forces are gravitational


er
tin

In accordance with (10.45),


on

.O

⎡ ⎤
C

0
not ⎢ ⎥
©

φ = gz b = −∇φ ≡ ⎣ 0 ⎦ (10.61)
−g
Introducing (10.60) and (10.61) in Bernouilli’s trinomial (10.58) yields
p 1 p 1 v2 de f
+ g z + v2 = const. =⇒ z + + = H = const. ∀x ∈ Γ
ρ0 2 ρ0 g 2 g
(10.62)
The terms in (10.61) have dimensions of length (height) and may be inter-
preted in the following manner.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Fluid Dynamics: Barotropic Perfect Fluids 471

Bernoulli’s theorem
p 1 v2 de f (10.63)
z + + = H = const. ∀x ∈ Γ
ρ0 g 2 g





elevation pressure velocity total
head head height

rs
Remark 10.4. The expression in (10.63) constitutes the so-called

ee
Bernoulli’s theorem (for an incompressible perfect fluid under

s gin
gravitational forces and in steady-state regime), which establishes
that the sum of the elevation, the pressure head and the velocity head

t d le En
is constant in every point belonging to a same streamline (see Fig-

r
ure 10.10).

ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

Remark 10.5. In engineering, water is generally considered an in-


compressible and perfect fluid, and the science that studies it is
le
liv or ec

named hydraulics. Since, in general, the body forces are of the gravi-
M

.A

tational type, Bernoulli’s Theorem is generally applicable in the res-


olution of steady-state problems in hydraulics.
m

d
uu
e
X Th

er
tin
on

.O
C

velocity head

pressure head

elevation

streamline

Figure 10.10: Physical interpretation of Benoulli’s theorem.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
472 C HAPTER 10. F LUID M ECHANICS

Example 10.2 – Determine the velocity of the water exiting the tank through
a small lateral hole placed at a distance h below the top surface of the water.
Consider the top of the tank is open and neglect the atmospheric pressure.
Assume a steady-state regime.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
Solution
C d P cs
b
a
i
The fluid in this problem (water) is an incompressible perfect fluid in steady-
an an n

state regime under gravitational forces and, thus, Bernoulli’s theorem can be
y ha

applied.
le
liv or ec

Consider a streamline originating at point A of the water surface and ending


M

.A

at point B of the exit hole (shown in the figure above). Applying Bernoulli’s
theorem between points A and B, and taking into account that the velocity
m

of the free surface in the tank is practically null and that its cross-section is
d
uu

much larger than that of the exit hole, then


e
X Th

er
tin

pA 1 v2 pB 1 v2B
zA + + A = zB + +

ρ0 g 2 g 
ρ0 g 2 g
on

.O

= h 

= 0 

=0 =0 =0
C

1 v2 √
h+0+0 = 0+0+ =⇒ v= 2gh .
2 g

10.3.4 Solution in Transient Regime


In a transient regime, the mechanical variables (in their spatial description) are
time-dependent (see Figure 10.8). The starting point to solve the problem is the
balance of linear momentum (10.54),

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Fluid Dynamics: Barotropic Perfect Fluids 473

 
1 ∂v
−∇ P + φ + v2 = ω ×v .
+ 2ω (10.64)
2 ∂t

In some cases, the solution to this equation in transient regime is particularly


simple. In the following subsections, several of these cases will be studied.

10.3.4.1 (Irrotational) Potential Flow


Consider the case of
• a perfect fluid

rs
• with potential body forces

ee
• and irrotational flow.

s gin
t d le En
Definition 10.4. The motion (or flow) of a fluid is said to be irrota-
tional (or potential) if the rotational of the velocity field is null at any

r
ba
ge ro or
point of this fluid.

eS m
ci
f

ra
C d P cs
b
a
i
In other words, an irrotational flow has a null vorticity vector.
an an n
y ha


le
⎨ ∇ × v (x,t) = 0
liv or ec

Irrotational flow ∀x ∀t (10.65)


M

⎩ ω (x,t) = 1 ∇ × v (x,t) = 0
.A

2
m

d
uu

If the flow is irrotational, it is inferred from (10.65) that there exists a scalar
e
X Th

function (denoted as velocity potential χ (x,t)) that satisfies6


er
tin

v (x,t) = ∇χ (x,t) . (10.66)


on

.O

Note that, in this case, the vector field v (x,t) is determined in terms of the scalar
C

velocity potential χ (x,t) (which becomes the main unknown of the problem).
©

Replacing the conditions (10.65) and (10.66) in (10.64) yields


   
1 ∂v ∂v ∂  ∂χ
− ∇ P + φ + v2 = + 2 ω ×v = = ∇χ (x,t) = ∇ =⇒
2 ∂t 
∂t ∂t ∂t
=0
(10.67)

6 It can be proven that, given an irrotational vector field v (x,t), that is, a vector field
that satisfies ∇ × v = 0, there exists a scalar function χ (x,t) (potential function) such that
v = ∇χ (x,t). Obviously, since ∇ × ∇ (•) ≡ 0, then ∇ × v = ∇ × ∇χ (x,t) = 0 is satisfied.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
474 C HAPTER 10. F LUID M ECHANICS

 
1 2 ∂χ
∇ P+φ + v + = ∇M (x,t) = 0
2 ∂t


M (x,t) ∀x ∀t (10.68)
∂ M (x,t)
=0 i ∈ {1, 2, 3}
∂ xi
This equation can be trivially integrated, resulting in
1 ∂χ
M (x,t) = P + φ + v2 + = ϕ (t) .

rs
(10.69)
2 ∂t

ee
Defining a modified velocity potential χ̄ (x,t) of the form

s gin

t ⎪
⎨ ∇χ̄ = ∇χ = v (x,t)

t d le En
de f
χ̄ (x,t) = χ (x,t) − ϕ (τ) dτ ⇒ ∂ χ̄ ∂χ (10.70)

r

⎩ = − ϕ (t)

ba
ge ro or
eS m
0 ∂t ∂t

ci
f

ra
C d P cs
and replacing (10.70) in (10.69) produces
b
a
i
an an n

1 ∂χ 1 ∂ χ̄
y ha

P + φ + v2 + − ϕ (t) = 0 =⇒ P + φ + (∇χ̄)2 + =0 ∀x ∀t ,
2 ∂t
2 ∂t
le
liv or ec

∂ χ̄
M

.A

∂t
(10.71)
m

which is the differential equation of hydraulic transients.


uu
e

The mechanical problem is then defined by:


X Th

er
tin

1) Mass continuity equation


on

.O

dρ dρ dρ
+ ρ∇ · v = + ρ ∇ · (∇χ̄) = 0 =⇒ + ρ∇2 χ̄ = 0 (10.72)
C

dt dt 
dt
©

∇2 χ̄

2) Balance of linear momentum (hydraulic transients equation)

1 ∂ χ̄
P (ρ, p) + φ + (∇χ̄)2 + =0 ∀x ∀t (10.73)
2 ∂t

3) Kinetic equation of state


ρ = ρ (p) (10.74)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Fluid Dynamics: Barotropic Perfect Fluids 475

These constitute a system of 3 scalar equations and 3 unknowns (p (x,t), ρ (x,t)


and χ̄ (x,t)) that can be integrated in the R3 × R+ domain. Once the potential
χ̄ (x,t) is known, the velocity field is obtained through

v (x,t) = ∇χ̄ (x,t) . (10.75)

10.3.4.2 Incompressible and Potential Flow


Consider the case of
• a perfect fluid

rs
• with potential body forces,
• irrotational flow

ee
• and incompressible flow.

s gin
Since the flow is incompressible, (10.46) and (10.72) allow determining7

t d le En

⎪ p

r
⎪ 1 p
⎨ P (p) =

ba
d p̄ =

ge ro or
eS m
dρ ρ ( p̄) ρ0
= 0 =⇒ ρ = ρ0 =⇒

ci
(10.76)
dt ⎪

0
f

ra

C d P cs
⎩ ∇2 χ̄ = Δ χ̄ = 0
not

b
a
i
an an n
y ha

and the mechanical problem (10.72) to (10.74) is reduced to:


le
liv or ec

1) Mass continuity equation


M

.A

∂ 2 χ̄
Δ χ̄ = =0 (10.77)
m

∂ xi ∂ xi
d
uu
e
X Th

er
tin

2) Balance of linear momentum (hydraulic transients equation)


on

.O

p 1 ∂ χ̄
+ φ + (∇χ̄)2 + =0 ∀x ∀t (10.78)
C

ρ0 2 ∂t
©

These constitute a system of 2 scalar equations and 2 unknowns (p (x,t) and


χ̄ (x,t)) that can be integrated in the R3 × R+ domain. In a steady-state regime,
the term ∂ χ̄/∂t = 0 and any time derivative in the system disappears, such that
the problem can be integrated in R3 .

7 Here, the differential operator named Laplace operator or Laplacian of (•) is defined as
not
Δ (•) = ∇ · ∇ (•) = ∇2 (•) = ∂ 2 (•)/∂ xi ∂ xi .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
476 C HAPTER 10. F LUID M ECHANICS

10.4 Fluid Dynamics: (Newtonian) Viscous Fluids


Consider now the general problem described by (10.1) to (10.9),


+ ρ∇ · v = 0 Mass continuity equation (1 eqn.)
dt
dv
∇ · σ + ρb = ρ Balance of linear momentum (3 eqns.)
dt
du
ρ = σ : d+ρ r−∇·q Energy balance (1 eqn.)

rs
dt

ee
σ = −p1 + λ Tr (d) 1 + 2μ d Mechanical constitutive equation (6 eqns.)

s gin
t d le En
s = s (d, θ , ρ) Entropy constitutive equation (1 eqn.)

r
ba
ge ro or
eS m
q = −K∇θ

ci
Heat conduction equation (3 eqns.)
f

ra
C d P cs
b
a
i
u = u (ρ, θ ) Caloric equation of state (1 eqn.)
an an n
y ha

le
F (ρ, p, θ ) = 0
liv or ec

Kinetic equation of state (1 eqn.)


M

.A

(10.79)
which constitute a system of 17 equations and 17 unknowns. This system is too
m

large to be treated efficiently and a reduced system of equations that allows a


d
uu

simpler resolution will be sought.


e
X Th

er
tin

10.4.1 Navier-Stokes Equation


on

.O

The Navier-Stokes equation is essentially the balance of linear momentum


C

of (10.79) expressed solely in terms of the velocity field v (x,t) and the pres-
©

sure p (x,t).

Lemma 10.3. The divergence of the strain rate tensor d (x,t) is re-
lated to the velocity field v (x,t) by
1 1
∇ · d = Δ v + ∇ (∇ · v) .
2 2

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Fluid Dynamics: (Newtonian) Viscous Fluids 477

Proof
  
  ∂ ∂ 1 ∂ vi ∂ v j 1 ∂ 2 vi 1 ∂ 2v j
∇·d = di j = + = + =
j ∂ xi ∂ xi 2 ∂ x j ∂ xi 2 ∂ xi ∂ x j 2 ∂ xi ∂ xi

1 ∂ ∂ vi 1 ∂ 2 v j 1 ∂ 1
= + = (∇ · v) + Δ v j =
2 ∂ x j ∂ xi 2 ∂ xi ∂ xi 2 ∂ x j 2 





Δvj [Δ v] j
∇·v [∇ (∇ · v)] j

1 1
= Δ v + ∇ (∇ · v) j ∈ {1, 2, 3}

rs
2 2 j

ee
(10.80)

s gin
1 1
∇ · d = Δ v + ∇ (∇ · v) (10.81)
2 2

t d le En

r
ba
ge ro or
eS m
ci
f
Lemma 10.4. Given a scalar function α (x,t), the following is sat-

ra
C d P cs
b
isfied.

a
i
∇ · (α 1) = ∇α
an an n
y ha

le
liv or ec

Proof
M

.A

  ∂ (αδi j ) ∂α ∂α  
m

∇ · (α 1) = = δi j = = ∇α i i ∈ {1, 2, 3} (10.82)
d

i ∂xj ∂xj ∂ xi
uu
e
X Th

er
tin

∇ · (α 1) = ∇α (10.83)
on

.O
C

Replacing the mechanical constitutive equation of (10.79) into the balance of


©

linear momentum of (10.79), and taking into account (10.81) and (10.83) leads
to

σ = −p1 + λ Tr (d) 1 + 2μ d ⎬
dv =⇒
∇ · σ + ρb = ρ ⎭
dt


⎪ ∇ · σ = −∇p + λ ∇ (Tr (d)) +μΔ v + μ∇ (∇ · v) (10.84)

⎨ 

∇ (∇ · v)

⎪ dv

⎩ ∇ · σ + ρb = −∇p + (λ + μ) ∇ (∇ · v) + μΔ v + ρb = ρ
dt

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
478 C HAPTER 10. F LUID M ECHANICS

which results in the Navier-Stokes equation.

Navier-Stokes equation
dv
−∇p + (λ + μ) ∇ (∇ · v) + μΔ v + ρb = ρ
dt (10.85)
∂p ∂ 2v
j ∂ 2v i dvi
− + (λ + μ) +μ + ρbi = ρ ; i ∈ {1, 2, 3}
∂ xi ∂ xi ∂ x j ∂ x j∂ x j dt

rs
ee
s gin
10.4.2 Energy Equation
The aim is to eliminate σ and q from the energy balance of (10.79) by replacing

t d le En
in this equation the mechanical constitutive equation and the entropy equation

r
of (10.79). To this aim, the definition of stress power in a Newtonian fluid (see

ba
ge ro or
eS m
Chapter 9) is recovered,

ci
f

ra
C d P cs
σ : d = WR + 2WD = −p ∇ · v + K Tr2 (d) + 2μ d : d ,
b
(10.86)

a
i
an an n

where d is the deviatoric part of the strain rate tensor. Fourier’s law is also
y ha

recovered,
le
liv or ec

q = −K ∇θ =⇒ ∇ · q = −∇ · (K ∇θ ) . (10.87)
M

.A

Replacing now in the energy balance of (10.79) yields


m

du
d

ρ = σ : d+ρ r−∇·q =⇒ (10.88)


uu
e

dt
X Th

er
tin

Energy equation
on

.O

du
= −p∇ · v + ρ r + ∇ · (K ∇θ ) + K Tr2 (d) + 2μ d : d
C

ρ
dt 

©

(10.89)
2WD
   
du ∂ vi ∂ ∂θ ∂ vi 2
ρ = −p +ρ r+ K +K + 2μ di j di j
dt ∂ xi ∂ xi ∂ xi ∂ xi

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Fluid Dynamics: (Newtonian) Viscous Fluids 479

10.4.3 Governing Equations of the Fluid Mechanics Problem


Considering the simplified versions of the balance of linear momentum (Navier-
Stokes equation (10.85)) and the energy balance (energy equation (10.89)) the
problem defined in (10.79) can be reduced to the following system of 7 PDEs
and 7 unknowns (ρ (x,t), v (x,t), p (x,t), u (x,t), θ (x,t)), which must be solved
in the R3 × R+ domain.

+ ρ∇ · v = 0 Mass continuity equation (1 eqn.)
dt

rs
−∇p + (λ + μ) ∇ (∇ · v) + Balance of linear momentum
(3 eqns.)

ee
dv (Navier-Stokes)
+μΔ v + ρb = ρ

s gin
dt
du
ρ = −p∇ · v + ρ r + ∇ · (K ∇θ ) +

t d le En
dt Energy balance (1 eqn.)
+K Tr2 (d) + 2μ d : d

r
ba
ge ro or
eS m
ci
u = u (ρ, θ ) f

ra
Caloric equation of state (1 eqn.)
C d P cs
b
a
i
an an n
y ha

F (ρ, p, θ ) = 0 Kinetic equation of state (1 eqn.)


le
liv or ec

(10.90)
M

.A

In the particular case of a barotropic regime (ρ = ρ (p)), the mechanical part


m

can be uncoupled from the thermal part in the set of equations of (10.79), result-
uu
e

ing in the mechanical problem defined by the following system of 5 equations


X Th

er

and 5 unknowns (ρ (x,t), v (x,t), p (x,t)).


tin
on


.O

+ ρ∇ · v = 0 Mass continuity equation (1 eqn.)


dt
C

−∇p + (λ + μ) ∇ (∇ · v) + Balance of linear momentum


dv (3 eqns.)
+μΔ v + ρb = ρ (Navier-Stokes)
dt
Kinetic equation of state
ρ = ρ (p) (1 eqn.)
(barotropic)
(10.91)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
480 C HAPTER 10. F LUID M ECHANICS

10.4.4 Physical Interpretation of the Navier-Stokes and Energy


Equations
Each of the terms in the Navier-Stokes equation (10.85),

⎪ dv

⎪ −∇p + (λ + μ) ∇ (∇ · v) + μΔ v + ρb − ρ =0

⎪ dt
⎨ 

a
∂p ! " (10.92)



⎪ − ∂ x + (λ + μ) ∇ (∇ · v) + μΔ v i + ρ bi − ρ ai = 0

⎩ i
i ∈ {1, 2, 3}

rs
ee
can be interpreted as a component of the system of forces (per unit of volume)

s gin
that acts on a volume differential of the fluid in motion as follows.

 

t d le En
−∇p − − (λ + μ) ∇ (∇ · v) + μΔ v + ρb + ρa = 0




r


ba
ge ro or
(10.93)

eS m
forces due to viscous forces exerted by body inertial

ci
the pressure forces forces
f
the contact between particles

ra
gradient
C d P cs
(= 0 when λ = μ = 0)
b
a
i
an an n

Figure 10.11 shows the projection of each of these components in the xi -


y ha

direction.
le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 10.11: Projection of the components of the Navier-Stokes equation in the xi -


direction.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Fluid Dynamics: (Newtonian) Viscous Fluids 481

Each of the terms in the energy equation (10.89) can also be given a physical
interpretation, as indicated in Table 10.1.

du
ρ = −p∇ · v + ρ r + ∇ · (k ∇θ ) + K Tr2 (d) + 2μ d : d
dt 

2WD

du variation of internal energy


ρ =

rs
dt unit of volume and unit of time

ee
d (dV )
variation of volume

s gin
∇ · v = dt =
dV unit of volume and unit of time

t d le En
d (dV ) mechanical work of the thermodynamic pressure
p =

r
−p∇ · v = − dt unit of volume and unit of time

ba
ge ro or
dV

eS m
(see Figure 10.12 and footnote8 )

ci
f

ra
C d P cs
heat generated by the internal sources and conduction
b
a
ρ r + ∇ · (k ∇θ ) =
i
unit of volume and unit of time
an an n
y ha

mechanical work of the viscous forces


2WD = σ D : d = dissipative
power =
le
unit of volume and unit of time
liv or ec
M

.A

Table 10.1: Physical interpretation of the energy equation.


m

d
uu
e
X Th

er
tin
on

.O
C

Figure 10.12: Mechanical work of the thermodynamic pressure.

8 Here, the relation d (dV )/dt = (∇ · v) dV is used (see Section 2.14.3 in Chapter 2).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
482 C HAPTER 10. F LUID M ECHANICS

10.4.5 Reduction of the General Problem to Particular Cases


The governing equations in fluid mechanics (10.90) can be simplified for certain
cases which are of particular interest in engineering applications.

10.4.5.1 Incompressible Fluids


In this case,

dρ ⎪
⎪ 
=0 ⎬ ρ = ρ0 = const.
dt =⇒ (10.94)

rs
dρ ⎪
⎪ ∇ · v = Tr (d) = 0
+ ρ∇ · v = 0 ⎭

ee
dt

s gin
and introducing (10.94) in (10.90) results in the governing equations detailed in
Table 10.2.

t d le En

r
ba
ge ro or
eS m
∇·v = 0

ci
Mass continuity equation
Mechanical
f

ra
C d P cs
Problem
b
a
dv
−∇p + μ Δ v + ρ0 b = ρ0
i
Navier-Stokes equation
an an n

dt
y ha

du
le
Energy balance ρ0 = ρ0 r + ∇ · (K ∇θ ) + 2μ d : d
liv or ec

Thermal dt
Problem
M

.A

Caloric equation of state u = u (ρ0 , θ )


m

d
uu

Constitutive equation σ = −p 1 + 2μ d
e
X Th

er
tin
on

Table 10.2: Governing equations in incompressible Newtonian fluids


.O
C

10.4.5.2 Fluids with Null Bulk Viscosity (Stokes Fluids)


In this case,
2 2 1
K=λ+ μ =0 =⇒ λ =− μ =⇒ λ +μ = μ (10.95)
3 3 3
2WD = K Tr2 (d) + 2μ d : d = 2μ d : d (10.96)


=0
and replacing (10.95) and (10.96) in (10.90) yields the governing equations
given in Table 10.3.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Fluid Dynamics: (Newtonian) Viscous Fluids 483


Mass continuity equation + ρ∇ · v = 0
dt

1 dv
Navier-Stokes equation −∇p + μ∇ (∇ · v) + μΔ v + ρb = ρ
3 dt

du
Energy balance ρ = −p∇ · v + ρ r + ∇ · (K ∇θ ) + 2μ d : d
dt

Caloric equation of state u = u (ρ, θ )

rs
ee
Kinetic equation of state F (ρ, p, θ ) = 0

s gin
2
Constitutive equation σ = −p 1 − μ Tr (d) 1 + 2μ d

t d le En
3

r
ba
ge ro or
eS m
ci
Table 10.3: Governing equations in Stokes fluids.
f

ra
C d P cs
b
a
i
an an n

10.4.5.3 Perfect Fluids


y ha

Perfect fluids have null viscosity, λ = μ = K = 0, and no heat conductivity,


le
liv or ec

K = 0. Introducing these conditions in (10.90) results in the problem shown in


M

.A

Table 10.4.
m

d
uu


e

Mass continuity equation + ρ∇ · v = 0


X Th

dt
er
tin

dv
on

.O

Euler’s equation −∇p + ρb = ρ


dt
C

du
©

Energy balance ρ = −p∇ · v + ρ r


dt

Caloric equation of state u = u (ρ, θ )

Kinetic equation of state F (ρ, p, θ ) = 0

Constitutive equation σ = −p 1

Table 10.4: Governing equations in perfect fluids.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
484 C HAPTER 10. F LUID M ECHANICS

10.4.5.4 Hydrostatics
In this case, the following conditions apply (see (10.20)):
dv
a= =0, ∇ · v = 0 , ρ = ρ0 , p = p0 and σ = −p0 1 , (10.97)
dt
and, thus, (10.90) is reduced to the equations described in Table 10.5.

Mechanical Hydrostatics fundamental equation −∇p0 + ρ0 b = 0

rs
Problem

ee
du
ρ0 = ρ0 r + ∇ · (K ∇θ )

s gin
Energy balance
Thermal dt
Problem

t d le En
Caloric equation of state u = u (ρ0 , θ )

r
ba
ge ro or
σ = −p0 1

eS m
Constitutive equation

ci
f

ra
C d P cs
b
a
i
Table 10.5: Governing equations in hydrostatics.
an an n
y ha

le
10.5 Boundary Conditions in Fluid Mechanics
liv or ec
M

.A

The governing equations of the fluid mechanics problem presented in the previ-
ous sections require adequate boundary conditions to be solved correctly. In gen-
m

eral, the spatial (or Eulerian) description is used in fluid mechanics problems,
d
uu

and a specific control volume (fixed in space) is analyzed, on whose boundary


e
X Th

er

the aforementioned spatial boundary conditions are applied. Even though there
tin

are different boundary conditions, and these often depend on the type of problem
on

being studied, the most common types of boundary conditions are summarized
.O

below.
C

10.5.1 Velocity Boundary Conditions


a) Prescribed velocity
In certain parts Γv̄ of the boundary of the control volume V being analyzed,
the velocities are known (see Figure 10.13).

v (x,t) = v̄ (x,t) ∀x ∈ Γv (10.98)

b) Impermeability condition
Usually, part of the boundary of the control volume V is composed of imper-
meable walls, Γvn , which are assumed to be impervious to fluid, that is, they

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Boundary Conditions in Fluid Mechanics 485

rs
Figure 10.13: Velocity boundary conditions: prescribed velocity.

ee
s gin
cannot be penetrated by said fluid. The mathematical expression of this con-
dition is denoted as impermeability condition and it establishes that the rela-

t d le En
tive velocity of the fluid, vr , with respect to the impermeable wall (assumed
mobile and with a velocity v∗ ) in the direction normal to the boundary must

r
ba
ge ro or
eS m
be null (see Figure10.14),

ci
f

ra
vn (x,t) = v · n = v∗ · n ∀x ∈ Γvn =⇒
C d P cs
b
a



i
an an n

fluid wall
(10.99)
y ha

vr · n = (v − v∗ ) · n = 0 .
le
∀x ∈ Γvn
liv or ec
M

.A

In the particular case of a fixed boundary, this condition is reduced to


(v∗ = 0) ⇒ v · n = 0 ∀x ∈ Γvn .
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 10.14: Velocity boundary conditions: impermeability condition.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
486 C HAPTER 10. F LUID M ECHANICS

Remark 10.6. The impermeability condition is usually applied for


perfect fluids (fluids without viscosity) in which the tangential com-
ponent of the relative velocity between the fluid and the wall vt (see
Figure 10.14) is assumed to be non-null.

c) Adherence condition

rs
In viscous fluids in contact with an impermeable wall, due to the effect of

ee
viscosity, the fluid is assumed to adhere to the wall (see Figure 10.15) and,

s gin
thus, the relative velocity between the fluid and the wall vr is null.

t d le En
vr (x,t) = v − v∗ = 0 ∀x ∈ Γv =⇒ v = v∗ ∀x ∈ Γv (10.100)

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

Figure 10.15: Velocity boundary conditions: adherence condition.


er
tin
on

.O

10.5.2 Pressure Boundary Conditions


C

In certain parts Γσ of the boundary, the traction vector t = σ · n can be pre-


scribed (see Figure 10.16).

t (x,t) = σ · n = t∗ (x,t) ∀x ∈ Γσ (10.101)

Under certain circumstances, only a part of the traction vector such as the ther-
modynamic pressure is prescribed. In effect, for a Newtonian fluid,

σ = −p1 + λ Tr (d) 1 + 2μ d =⇒
(10.102)
t = σ · n = −p n + λ Tr (d) n + 2μ d · n ,

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Boundary Conditions in Fluid Mechanics 487

Figure 10.16: Pressure boundary conditions: prescribed traction vector.

which exposes how the thermodynamic pressure p is a part of the normal com-

rs
ponent of the traction vector t. The prescription of the thermodynamic pressure
on a part of the boundary Γp is written as

ee
s gin
p (x,t) = p∗ (x,t) ∀x ∈ Γp . (10.103)

t d le En

r
ba
ge ro or
eS m
10.5.3 Mixed Boundary Conditions

ci
f

ra
C d P cs
In certain cases (such as the entrance and exit sections of pipes) the pressure (a
b
a
part of the normal component of the traction vector) and the tangential com-
i
an an n

ponents of the velocity (which are assumed to be null, see Figure 10.17) are
y ha

prescribed.
le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

Figure 10.17: Mixed boundary conditions.

10.5.4 Boundary Conditions on Free Surfaces

Definition 10.5. A free surface is a contact surface between the air


(atmosphere) and a fluid (generally water).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
488 C HAPTER 10. F LUID M ECHANICS

free surface

= free surface height

Figure 10.18: Free surface of the sea.

rs
ee
Examples of free surface9 are the surface of the sea (see Figure 10.18) or the

s gin
surface that separates the saturated and unsaturated parts of an embankment dam
(see Figure 10.19).

t d le En
A hypothesis with a clear physical sense that is frequently used in relation

r
to a free surface is that such a surface is a material surface (constituted always

ba
ge ro or
eS m
by the same particles). This hypothesis implicitly establishes certain boundary

ci
f
conditions on the velocity field in the material surface Γf s . In effect, considering

ra
C d P cs
b
the free surface in Figure 10.18,

a
i
an an n

Γf s := {x | φ (x, y, z,t) ≡ z − η (x, y,t) = 0} , (10.104)


y ha

le
liv or ec

and imposing the material character of the free surface (null material derivative,
see Section 1.11 in Chapter 1),
M

.A

dφ ∂φ ∂η ∂η ∂η ∂φ
m

= + v · ∇φ = − − vx − vy + vz =0, (10.105)
d

dt ∂t ∂t ∂x ∂y ∂z
uu



e
X Th

=1
er
tin

∂η ∂η ∂η
on

.O

vz (x,t) = + vx + vy ∀x ∈ Γf s . (10.106)
∂t ∂x ∂y
C

This condition establishes the dependency of the vertical component of the ve-
locity vz on the other components vx and vy .
Another boundary condition frequently imposed on free surfaces is that, in
these surfaces, the thermodynamic pressure is known and equal to the atmo-
spheric pressure10 ,
p (x,t) = Patm ∀x ∈ Γf s (10.107)

9 In general, in fluid mechanics problems in which free surfaces appear, the position of these
surfaces is not known and their geometrical characteristics become an unknown of the prob-
lem.
10 The value of the atmospheric pressure is generally neglected (P
atm ≈ 0).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Laminar and Turbulent Flows 489

dry free surface

saturated

rs
ee
s gin
Figure 10.19: Free surface of an embankment dam.

t d le En
Equation (10.107) allows identifying, in certain cases, the position of the free

r
ba
ge ro or
surface (once the pressure field is known) as the locus of points in the fluid in

eS m
ci
which the pressure is equal to the atmospheric pressure.
f

ra
C d P cs
b
a
i
an an n

Equation of the Γf s := {x | p (x,t) − Patm = 0} (10.108)


y ha

free surface
le
liv or ec
M

.A

10.6 Laminar and Turbulent Flows


m

d
uu
e

10.6.1 Laminar Flow


X Th

er
tin

The equations governing a fluid mechanics problem, described in the previous


sections, are valid for a certain range of motion of the fluids, named laminar
on

.O

flow (or regime). Basically, laminar flow is physically characterized by the fact
C

that the fluid moves in parallel layers that do not mix (see Figure 10.20).
©

Vortex

Figure 10.20: Laminar flow around an obstacle.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
490 C HAPTER 10. F LUID M ECHANICS

The character of a laminar flow is identified by the Reynolds number Re

de f V ×L
Reynolds number: Re =
ν

⎨ V = characteristic velocity of the fluid
⎪ (10.109)
L = characteristic length of the domain


ν = kinematic viscosity (ν = μ/ρ)

rs
such that small values of the Reynolds number characterize laminar flows.

ee
10.6.2 Turbulent Flow

s gin
When the velocity increases and the viscosity decreases, the Reynolds num-

t d le En
ber (10.109) increases. For increasing values of this number, the initially lam-
inar flow is seen to distort and become highly unstable. The flow can then be

r
ba
ge ro or
understood as being in a situation in which the velocity v (x,t), at a given point

eS m
ci
in space, randomly and rapidly fluctuates along time about a mean value v̄ (x,t)
f

ra
C d P cs
(see Figure 10.21). This situation is defined as turbulent flow (or regime).
b
a
Even though the equations of the fluid mechanics problem in general, and the
i
an an n

Navier-Stokes equation in particular, are still valid in turbulent regime, certain


y ha

circumstances (such as the difficulty in treating the mathematical problem and


le
liv or ec

the impossibility of experimentally characterizing the rapid fluctuations of the


variables of this problem) impose a singular treatment for turbulent flow. The
M

.A

mathematical characterization of turbulent regime is done, then, by means of


the so-called turbulence models. These models are based on isolating the mean
m

values of the velocity and pressure fields from their fluctuations and, then, the
uu
e

governing equations of the problem are obtained in terms of these mean values.
X Th

er
tin
on

.O
C

Figure 10.21: Variation of the velocity along time in laminar and turbulent flows.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Fluid Mechanics Formulas 491

10.7 Fluid Mechanics Formulas


10.7.1 Stress tensor for Newtonian fluids
(incompressible fluid, ∇ · v = 0)

Cartesian coordinates
 
∂ vx ∂ vx ∂ vy
σx = 2μ −p τxy = τyx = μ +
∂x ∂y ∂x
 
∂ vy ∂ vy ∂ v z

rs
σy = 2μ −p τyz = τzy = μ + (10.110)
∂y ∂z ∂y

ee
 
∂ vz ∂ v z ∂ vx

s gin
σz = 2μ −p τzx = τxz = μ +
∂z ∂x ∂z

t d le En

r
Cylindrical coordinates

ba
ge ro or
eS m
 

ci
∂ vr f ∂  vθ  1 ∂ vr

ra
σr = 2μ −p τrθ = τθ r = μ r +
C d P cs
∂r ∂r r r ∂θ
b
a
   
i
an an n

1 ∂ v θ vr ∂ v θ 1 ∂ vz
σθ = 2μ + −p τθ z = τzθ = μ +
y ha

r ∂θ r ∂z r ∂θ
 
le
liv or ec

∂ vz ∂ vz ∂ vr
σz = 2μ −p τzr = τrz = μ +
M

.A

∂z ∂r ∂z
(10.111)
m

1 ∂ 1 ∂ vθ ∂ vz
d

∇·v = (rvr ) + + (10.112)


uu

r ∂r r ∂θ ∂z
e
X Th

er
tin

Spherical coordinates
on

.O

∂ vr
C

σr = 2μ −p
∂r
©

 
1 ∂ vθ vr
σθ = 2μ + −p (10.113)
r ∂θ r
 
1 ∂ vφ vr vθ cot θ
σφ = 2μ + + −p
r sin θ ∂ φ r r

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
492 C HAPTER 10. F LUID M ECHANICS

 
∂  v θ  1 ∂ vr
τrθ = τθ r = μ r +
∂r r r ∂θ
  
sin θ ∂ vφ  1 ∂ vθ
τ θ φ = τφ θ = μ + (10.113 (cont.))
r ∂ θ sin θ r sin θ ∂ φ
 
1 ∂ vr ∂ vφ 

τφ r = τrφ = μ +r
r sin θ ∂ φ ∂r r

1 ∂  2  1 ∂ 1 ∂ vφ
∇·v = r vr + (vθ sin θ ) + (10.114)
r ∂r
2 r sin θ ∂ θ r sin θ ∂ φ

rs
ee
s gin
10.7.2 Continuity Equation
Cartesian coordinates

t d le En
∂ρ ∂ ∂ ∂

r
+ (ρvx ) + (ρvy ) + (ρvz ) = 0

ba
(10.115)

ge ro or
eS m
∂t ∂ x ∂y ∂z

ci
f

ra
C d P cs
b
a
Cylindrical coordinates
i
an an n

∂ρ 1 ∂ 1 ∂ ∂
y ha

+ (ρrvr ) + (ρvθ ) + (ρvz ) = 0 (10.116)


∂t r ∂r r ∂θ ∂z
le
liv or ec
M

.A

Spherical coordinates
m

∂ρ 1 ∂  2  1 ∂ 1 ∂  
d

+ 2 ρr vr + (ρvθ sin θ ) + ρvφ = 0 (10.117)


uu

∂t r ∂ r r sin θ ∂ θ r sin θ ∂ φ
e
X Th

er
tin
on

.O

10.7.3 Navier-Stokes Equation


C

(incompressible fluid, ∇ · v = 0; ρ and μ const.)


©

Cartesian coordinates
 2   
∂p ∂ vx ∂ 2 vx ∂ 2 vx ∂ vx ∂ vx ∂ vx ∂ vx
− +μ + + + ρb x = ρ + v x + vy + vz
∂x ∂ x2 ∂ y2 ∂ z2 ∂t ∂x ∂y ∂z
 2   
∂p ∂ vy ∂ vy ∂ vy
2 2 ∂ vy ∂ vy ∂ vy ∂ vy
− +μ + + 2 + ρby = ρ + vx + vy + vz
∂y ∂ x2 ∂ y2 ∂z ∂t ∂x ∂y ∂z
 2   
∂p ∂ vz ∂ vz ∂ vz
2 2 ∂ vz ∂ vz ∂ vz ∂ vz
− +μ + + 2 + ρbz = ρ + vx + vy + vz
∂z ∂ x2 ∂ y2 ∂z ∂t ∂x ∂y ∂z
(10.118)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Fluid Mechanics Formulas 493

Cylindrical coordinates

   
∂p ∂ 1 ∂ 1 ∂ 2 vr 2 ∂ vθ ∂ 2 vr
− +μ (rvr ) + 2 − + + ρbr =
∂r ∂r r ∂r r ∂ θ 2 r2 ∂ θ ∂ z2
 
∂ vr ∂ vr vθ ∂ vr v2θ ∂ vr
=ρ + vr + − + vz
∂t ∂r r ∂θ r ∂z

   
1∂p ∂ 1 ∂ 1 ∂ 2 vθ 2 ∂ vr ∂ 2 vθ

rs
− +μ (rvθ ) + 2 + 2 + + ρbθ =
r ∂θ ∂r r ∂r r ∂θ2 r ∂θ ∂ z2

ee
 
∂ vθ ∂ vθ vθ ∂ vθ vr vθ ∂ vθ

s gin
=ρ + vr + + + vz
∂t ∂r r ∂θ r ∂z

t d le En
   
∂p 1 ∂ ∂ vz 1 ∂ 2 vz ∂ 2 vz

r
− +μ + 2 + 2 + ρbz =

ba
ge ro or
r

eS m
∂z r ∂r ∂r r ∂θ2 ∂z

ci
 
f

ra
∂ vz ∂ vz vθ ∂ vz ∂ vz
C d P cs
=ρ + vr + + vz
b
a
∂t ∂r r ∂θ ∂z
i
an an n

(10.119)
y ha

le
Spherical coordinates
liv or ec
M

.A

#    
∂p ∂ 1 ∂  2  1 ∂ ∂ vr 1 ∂ 2 vr
m

− +μ r v + sin θ + +
d

r
∂r ∂ r r2 ∂ r r2 sin θ ∂ θ ∂$θ r2 sin2 θ ∂ φ 2
uu
e

∂ 2 ∂ vφ
X Th

2
er
tin

− 2 (vθ sin θ ) − 2 + ρbr =


r sin θ ∂#θ r sin θ ∂ φ $
on

.O

vφ ∂ vr vθ + vφ
2 2
∂ vr ∂ vr vθ ∂ vr
=ρ + vr + + −
C

∂t ∂r r ∂ θ r sin θ ∂ φ r
©

#    
1∂p 1 ∂ ∂ vθ 1 ∂ 1 ∂
− +μ 2 r 2 + 2 (vθ sin θ ) +
r ∂θ r ∂r ∂r r ∂ θ sin θ ∂ θ $
1 ∂ 2 vθ 2 ∂ vr 2 cot θ ∂ vφ
+ 2 2 + 2 − + ρbθ =
#r sin θ ∂ φ r ∂ θ r2 sin θ ∂ φ
2
$
vφ ∂ vθ vr vθ vφ cot θ
2
∂ vθ ∂ vθ vθ ∂ vθ
=ρ + vr + + + −
∂t ∂r r ∂θ r sin θ ∂ φ r r
(10.120)

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
494 C HAPTER 10. F LUID M ECHANICS

#    
1 ∂p 1 ∂ ∂ vφ 1 ∂ 1 ∂  
− +μ 2 r2 + 2 vφ sin θ +
r sin θ ∂ φ r ∂r ∂r r ∂ θ sin θ ∂ θ $
1 ∂ vφ 2
2 ∂ vr 2 cot θ ∂ vθ
+ 2 2 + 2 + + ρbϕ =
r sin θ ∂ φ 2 r sin θ ∂ φ r2 sin θ ∂ φ
 
∂ vφ ∂ v φ v θ ∂ vφ vφ ∂ vφ vφ vr vθ vφ
=ρ + vr + + + + cot θ
∂t ∂r r ∂θ r sin θ ∂ φ r r
(10.120 (cont.))

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 495

P ROBLEMS

Problem 10.1 – The barotropic fluid flowing inside the pipe shown in the figure
below has the following kinetic equation of state.
 
ρ
p = β ln (β and ρ0 const.)
ρ0

rs
ee
Determine, for a steady-state regime, the exit pressure P2 in terms of the other

s gin
variables shown in the figure.

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec

Solution
M

.A

The global spatial form of the mass continuity equation (5.22) states
m


d

d
uu

ρ dV = 0 .
e

dt
X Th

er
tin

Using the Reynolds Transport Theorem (5.37) on this expression results in


on

.O

    
∂ ∂
C

d
ρ dV = ρ dV + ρv · n dS =⇒ ρ dV + ρv · n dS = 0 ,
©

dt ∂t ∂t
V V ∂V V ∂V

and introducing the conditions associated with a steady-state regime yields


 

ρ dV = 0 =⇒ ρv · n dS = 0 .
∂t
V ∂V

Applying this last expression to the problem described in the statement produces

−ρ1 v1 S1 + ρ2 v2 S2 = 0 =⇒ ρ1 v1 S1 = ρ2 v2 S2 .

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
496 C HAPTER 10. F LUID M ECHANICS

Finally, isolating the density from the given kinetic equation of state,
 
ρ
p = β ln =⇒ ρ = ρ0 e p/β ,
ρ0
and introducing it into the previous one produces
v 1 S1
ρ0 e P1 /β v1 S1 = ρ0 e P2 /β v2 S2 =⇒ e (P2 −P1 )/β = =⇒
v2 S2
 

rs
v1 S1
P2 = P1 + β ln .

ee
v2 S2

s gin
t d le En

r
Problem 10.2 – Determine the value per unit of thickness of the horizontal

ba
ge ro or
eS m
force F that must be applied on point B of the semicircular floodgate shown

ci
f
in the figure such that the straight line AB remains vertical. The floodgate can

ra
C d P cs
rotate around the hinge A and separates two different height levels, h and αh,
b
a
i
of a same fluid.
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O

H YPOTHESES :
C

1) The weight of the floodgate can be neglected.


2) The atmospheric pressure is negligible.

Solution
The only forces acting on the floodgate are the pressure forces of the fluids,
the force F and the reaction in A (horizontal component H and vertical compo-
nent V).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 497

Figure A Figure B
Since the pressure exerted by the fluids is perpendicular to the surface of the

rs
floodgate and the floodgate is circular, the resultant force obtained by integrating

ee
the pressures on the surface are applied at the center of the circumference defined

s gin
by the floodgate. Thus, posing the equilibrium of momentum with respect to the
center of this circumference (see Figure A) results in

t d le En
FR = HR =⇒ H=F.

r
ba
ge ro or
eS m
Imposing now the equilibrium of horizontal forces, knowing that the fluids exert

ci
f
a horizontal pressure with a triangular distribution (see Figure B), yields

ra
C d P cs
b
a
i
1 1 1  
an an n

2F + (ρgα h)(α h) = (ρgh) h =⇒ F = ρgh2 1 − α 2 .


y ha

2 2 4
le
liv or ec
M

.A
m

Problem 10.3 – Determine the relation between the. force F applied on the
d

piston shown in the figure and its velocity of descent δ .


uu
e
X Th

er
tin
on

.O
C

H YPOTHESES :
1) Assume the fluid is an incompressible perfect fluid in steady-state regime.
2) The atmospheric pressure is negligible.
3) S1 and S2 are the cross-sections.
4) The density of the fluid is ρ.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
498 C HAPTER 10. F LUID M ECHANICS

Solution
The stress state of a perfect fluid is known to be of the form σ = −p 1 (see
Section 9.3 in Chapter 9). The mass continuity equation (5.22) is applied to
obtain the relation between the velocities of the fluid,
S1 S1 .
v1 S1 = v2 S2 =⇒ v2 = v1 = δ .
S2 S2
Taking into account Bernoulli’s theorem (10.63) between an arbitrary point in
contact with the piston and another at the exit cross-section, both belonging to a

rs
same streamline, results in
#  $

ee
.  
p δ2 S1 . 2 1 ρ S1 2 .

s gin
H+ + = 0+0+ δ =⇒ p = − 1 δ 2 − ρg H
ρg 2g S2 2g 2 S2

t d le En
Therefore, p must be constant for any point in contact with the piston (x = H).

r
ba
ge ro or
Then,

eS m
p = const. ∀x = H =⇒ F = p S1 .

ci
f

ra
.
C d P cs
Finally, the force F is related to δ in the following manner.
b
a
i
an an n

# 2 $
y ha

ρ S1 .
F = S1 − 1 δ 2 − ρg H S1
le
liv or ec

2 S2
M

.A
m

d
uu
e

Problem 10.4 – A shear force f ∗ per unit of surface acts on an rigid plate of
X Th

er
tin

indefinite size with density ρ ∗ and thickness t. The plate slides at a velocity v∗
in the longitudinal direction on a plane inclined at an angle α with respect to
on

.O

the horizontal longitudinal direction. Between the plate and the inclined plane
C

are two distinct and immiscible Newtonian fluids with viscosities μ1 and μ2 ,
©

which are distributed into two layers with the same thickness h.
a) Establish the generic form of the pressure and velocity fields and argue the
hypotheses used to determine them.
b) Integrate the corresponding differential equations and obtain, except for the
integration constants, the distribution of pressures and velocities in each
fluid.
c) Indicate and justify the boundary conditions that must be applied to deter-
mine the above integration constants.
d) Completely determine the pressure and velocity fields as well as the stresses
in each fluid. Plot the distribution of each variable (velocities, pressure and

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 499

stresses) on a cross-section such as A − A , indicating the most significant


values.
e) Obtain the value of v∗ in terms of f ∗ and the volume flow rate q that flows
through a semicircular section such as B − B .

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
H YPOTHESES :
b
a
i
an an n

1) Both fluids are incompressible.


y ha

2) Assume a steady-state regime.


le
liv or ec

3) The body forces of the fluids can be neglected.


4) The atmospheric pressure is negligible.
M

.A
m

d
uu
e
X Th

Solution
er
tin

a) Note that the z-dimension, perpendicular to the plane of the paper, does not
on

.O

intervene in the problem. Thus, the pressure and velocity fields are reduced to
not  T
C

p = p (x, y) and v = v (x, y) ≡ vx (x, y) , 0 , 0 .


©

In fact, vx does not depend on x either since the velocity should be the same in
all the cross-sections of the type A − A . If this is not acknowledged a priori, the
mass continuity equation (5.22) may be imposed, considering the incompress-
ible nature of the fluids, as follows.
dρ ∂ vx ∂ vy
+ ρ∇ · v = 0 =⇒ ∇·v = 0 =⇒ + = 0 , but vy = 0
dt ∂x ∂y
∂ vx not  T
=⇒ =0 =⇒ vx = vx (y) =⇒ v ≡ vx (y) , 0 , 0
∂x

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
500 C HAPTER 10. F LUID M ECHANICS

Therefore, the pressure and velocity fields are


not  T
p = p (x, y) and v ≡ vx (y) , 0 , 0 .

b) The components of the Navier-Stokes differential equation (10.85) in Carte-


sian coordinates must be integrated to obtain the expressions of v and p,
∂p ∂ 2 vx
x − component =⇒ 0=− +μ 2 ,
∂x ∂y

rs
∂p

ee
y − component =⇒ 0=− ⇒ p = p (x) .
∂y

s gin
The pressure p only depends on x and the component vx of v only depends on y.

t d le En
Therefore, the partial derivatives in the equation for the x-component can be
replaced by total derivatives. In this way, an equality of functions is obtained in

r
ba
ge ro or
which the pressure term depends solely on x whilst the velocity term depends

eS m
ci
exclusively on y. Consequently, these terms must be constant.
f

ra

C d P cs
b
a
dp d 2 vx ⎪
=μ 2 ⎬
i
d 2 vx
an an n

dp
dx dy =⇒ =μ = k = const.
y ha


⎭ dx dy2
f (x) = f (y)
le
liv or ec
M

.A

dp
=k =⇒ p (x) = k x + A
dx
m

d 2 vx
uu

k k 2
e

= =⇒ vx (y) = y + By + C
X Th

dy 2 μ 2μ
er
tin

To determine the stresses, the constitutive equation in Cartesian coordinates of


on

.O

Table10.2 is used,

C

σx = σy = σz = −p (x) ⎬
©

∂ vx (y) =⇒
τxy = τyx = μ ⎭
∂y
⎡   ⎤
k
⎢ −k x − A μ μ
y+B 0 ⎥
⎢   ⎥
σ (x, y) ≡ ⎢ ⎥
not
⎢μ k ⎥
⎣ y+B −k x − A 0 ⎦
μ
0 0 −k x − A

where the constants in these expressions (k, A, B, C) are different for each fluid.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 501

c) The boundary conditions that must be applied in this problem are:

V ELOCITY BOUNDARY CONDITIONS



1. v1x (y)y=h = v∗ , since the plate moves at a velocity v∗ and μ > 0.

2. v2x (y)y=−h = 0 , since the inclined plane does not move and μ > 0.
 
3. v1x (y)y=0 = v2x (y)y=0 , which is the continuity condition for v at the inter-
face between the two fluids.

rs
P RESSURE BOUNDARY CONDITIONS

ee
In the fluid with density μ1 , the pressure is prescribed for y = h or, directly,

s gin
since p does not depend on y (because the weight of the fluid is neglected), the
pressure p1 is prescribed in the whole domain of this fluid. The value of p1

t d le En
corresponds to the pressure that the plate exerts on the fluid with density μ1 ,
which is the projection of the plate’s weight in the direction of the y-axis.

r
ba
ge ro or
eS m
4. W = ρ ∗ gt is the weight of a section of the plate with unit length, according

ci
f

ra
to the x- and z-axis. Here, patm = 0 has been considered.
C d P cs
b
a
5. p1 = ρ ∗ gt cos α , ∀x is the projection on the y-axis. Since a unit length has
i
an an n

been considered, the weight is directly the exerted pressure.


y ha

 
6. p1  = p2  ∀x is the continuity condition for the pressure in the
le
liv or ec

y=0 y=0
interface between the two fluids.
M

.A
m

S TRESS BOUNDARY CONDITIONS


d
uu
e

The continuity condition for stresses that must be imposed in the interface be-
X Th

tween the two fluids does not affect the complete tensor σ . Instead, only the
er
tin

traction vector t is affected. The condition


on

 
.O

t1  = − t2 
C

y=0 y=0
©

must be satisfied. Considering that the unit normal vector n is the exterior normal
vector, then
n1 ≡ [0 , −1 , 0]T n2 ≡ [0 , 1 , 0]T .
not not
and
Hence, the shear stresses must satisfy:
1
 2

7. τxy = τxy
y=0 y=0

d) Only 7 boundary conditions have been established and 8 constants must be


determined, but since some equations include two constants, it suffices. Replac-
ing the expressions of p, v and σ in the boundary conditions results in:

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
502 C HAPTER 10. F LUID M ECHANICS

k1 2
h + B1 h + C1 = v∗
2μ1
k2 2
h − B2 h + C2 = 0
2μ2
C1 = C2

k1 = 0
k1 x + A1 = ρ ∗ gt cos α , ∀x =⇒
A1 = ρ ∗ gt cos α


rs
k1 = k 2 = 0
k1 x + A1 = k2 x + A2 , ∀x =⇒

ee
A1 = A2 = ρ ∗ gt cos α

s gin
%
y=0
=⇒ μ1 B1 = μ2 B2

t d le En
k1 = k2 = 0

r
ba
ge ro or
Solving and replacing these values in the expressions for the pressure, velocity

eS m
ci
and stress obtained in b) results in
f

ra
C d P cs
b
a
 
i
v∗ y μ1
an an n

v1x (y) = μ1 + p1 = p2 = ρ ∗ gt cos α = const.


y ha

1+ h μ2
μ2
le
liv or ec

v∗
μ1 v∗ y  τxy
1 = τ2 = μ   = const.
M

1
.A

xy
v2x (y) = +1 μ1
μ2 1 + μ1 h h 1+
μ2
m

μ2
d
uu
e
X Th

er
tin

e) To determine the relation between f ∗ and v∗ , the equilibrium of forces on a


on

unit element of the plate is posed. Three forces act on this element:
.O

1) The force f ∗ that pushes the plate in the positive direction of the x-axis.
C

2) The projection of the plate’s own weight in the direction of the x-axis. This
force pulls the plate in the negative direction of the x-axis.
3) The shear force of the fluid on the plate, which resists the motion of the plate
and, thus, acts in the negative direction of the x-axis.
To determine the sign criterion of this last force, the stresses acting on an element
of the fluid domain are drawn:

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 503

rs
Posing the equilibrium of forces yields

ee
v∗

s gin
f ∗ = ρ ∗ gt sin α + μ1  
μ1
1+ h

t d le En
μ2

r
and, isolating v∗ , produces the velocity in terms of the shear force,

ba
ge ro or
eS m
ci
  f

ra
μ1
C d P cs
h
v∗
b
(f ∗ − ρ ∗ gt sin α)

a
= 1+ .
μ1 μ2
i
an an n
y ha

To compute the volume flow rate that flows across the surface B−B , one must
le
liv or ec

take into account that the fluids are incompressible and, thus, the volume flow
M

.A

rate crossing the curved surface is the same as if a straight segment joining B
and B was considered, that is,
m

d
uu

  h
e
X Th

q= v · n dS = v · n dS = vx (y) dy .
er
tin

BB curved BB straight −h
on

.O

Then, replacing the expressions found in d) for the velocity vx and integrating
C

results in the volume flow


©

 
1 μ1
q= v∗ h + .
2 μ1 + μ2

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
504 C HAPTER 10. F LUID M ECHANICS

Problem 10.5 – Figure A shows the cross-section of a damper of indefinite


length composed of a piston ABA B that slides inside a container filled with
an incompressible
. Newtonian fluid with viscosity μ. The piston descends at a
velocity δ (t), producing a lateral flow of fluid between the piston and the walls
(see Figure B).

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n

Figure A Figure B
y ha

le
liv or ec

a) Determine the pressure and velocity fields in the zone of the fluid shown in
Figure B (zone ABCD), except for the integration constants.
M

.A

b) Indicate and justify the boundary conditions that must be applied to deter-
m

mine the above integration constants.


d
uu

c) Completely determine the pressure and velocity fields in zone ABCD of the
e
X Th

fluid.
er
tin

d) Determine the expression of the stress tensor in zone ABCD of the fluid.
e) Assuming that the stress σy in the surface A−A is uniform and equal
on

.O

. to
the stress in point A, prove there exists a relation of the form F.= η δ (t),
C

where F is the force per unit of length applied on the piston and δ (t) is the
©

velocity of descent of said piston. Compute the value of η.


H YPOTHESES :
1) The body forces of the fluid can be neglected.
2) The weight of the piston can be neglected.
3) Assume a steady-state regime.
4) The atmospheric pressure is negligible.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 505

Solution
a) The problem is not defined in the z-direction, the direction perpendicular to
the plane of the paper, and, thus, is independent of the z variable. Then, consider
the bidimensional situation
not  T
v ≡ vx (x, y) , vy (x, y) , 0 .

On the other hand, vx = 0 must be satisfied on the


walls AB and CD, owing to the impermeability con-
dition (a fluid cannot penetrate through a solid).

rs
For convenience, an additional approximate hypoth-

ee
esis is introduced to further simplify the problem: it
will be assumed that vx = 0 in all the zone ABCD of

s gin
the fluid. However, the streamlines have, in fact, the

t d le En
approximate form shown in the figure to the right.
It is even possible that vortexes are formed in this

r
ba
region if there is a high velocity.

ge ro or
eS m
In short, the velocity and pressure fields are assumed to be of the form

ci
f

ra
not  T
C d P cs
v ≡ 0 , vy (x, y) , 0
b
and p = p (x, y) .

a
i
an an n

The mass continuity equation (5.22) for an incompressible fluid (ρ = const.) is


y ha

reduced to ∇ · v = 0 and, for this particular problem,


le
liv or ec

∂ vy
M

.A

=0 =⇒ vy = vy (x) .
∂y
m

d
uu

Then, the velocity remains constant for a same vertical line since the spatial
e
X Th

description of the velocity does not depend on y.


er
tin

Now, the Navier-Stokes equation (10.85) in Cartesian coordinates is imposed,


considering the hypotheses given in the statement of the problem and the ad-
on

.O

ditional assumptions described above. Since the problem is bidimensional, the


C

z-component of the equation does not provide information.


©


∂p
0=− =⇒ p = p (y) ⎪ ⎬ ∂p ∂ 2 vy
∂x =⇒ = μ
∂p ∂ vy
2

⎭ ∂y ∂ x2
0=− +μ 2
∂y ∂x
The term in the right-hand side of the equation depends solely on x and the one
in the left-hand side depends only on y, therefore both terms must be constant.
∂p
=k =⇒ p = ky +C1
∂y

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
506 C HAPTER 10. F LUID M ECHANICS

∂ 2 vy ∂ vy k 1k 2
μ =k =⇒ = x +C2 =⇒ vy (x) = x +C2 x +C3
∂ x2 ∂x μ 2μ

b) The boundary conditions that must be applied in this problem are:

V ELOCITY BOUNDARY CONDITIONS




1. vy (x)  = 0 , ∀y, since there is no relative displacement of the fluid with
x=0
respect to the wall.


rs
 .
2. vy (x)  = −δ , ∀y, again, since there is no relative displacement.

ee
x=a

s gin
P RESSURE BOUNDARY CONDITIONS

t d le En


3. p (y)  = patm = 0

r
ba
ge ro or
y=m+h

eS m
ci
f

ra
C d P cs
VOLUME FLOW RATE BOUNDARY CONDITIONS
b
a
i
an an n

In an incompressible fluid the entrance and exit volume flow rates are the same,
y ha

Qin = Qout , where 


le
Q= v · n dS .
liv or ec
M

.A

S
.
The piston descends at a velocity δ and, thus, its cross-section is introduced into
m

the fluid, pushing it upwards. Then, the entrance volume flow rate can be defined
uu
e

as (velocity · surface),
X Th

.
er
tin

Qin = δ · L .
on

.O

On the other hand, the exit volume flow rate, flowing in the space left between
the piston and the lateral walls, is determined by means of the general expression
C

for volume flow rate


©

 a
Qout = 2 v · n dS = 2 vy (x) dx .
Sa 0

Finally, equating the entrance and exit volume flow rates results in:
a .
4. 2 vy (x) dx = δ L
0

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 507

c) The constants are determined by means of the boundary conditions described


in b) as follows:


vy (x)  = 0 =⇒ vy (0) = C3 =⇒ C3 = 0
x=0

 .
 . 1k 2 δ k
vy (x)  = −δ =⇒ vy (a) = a +C2 a =⇒ C2 = − − a
x=a 2μ a 2μ
a a    

rs
k 2 k a3 a2 .
2 vy (x) dx = 2 x +C2 x dx = 2 +C2 = δL =⇒

ee
2μ 2μ 3 2

s gin
0 0
. 
6μ . δ L

t d le En
k = − 3 δ (a + L) and C2 = 2+3
a a a

r
ba
ge ro or
eS m
ci
 f

ra
 6μ .
C d P cs
p (y)  =0 =⇒ k (m + h)+C1 = 0 =⇒
b C1 = δ (a + L) (m + h)

a
a3
i
y=m+h
an an n
y ha

le
Introducing these values in the expressions for the pressure and velocity obtained
liv or ec

in a) results in:
M

.A
m

6μ .
d

p = p (y) = δ (a + L) (m + h − y)
uu

a3
e

. 2 δ 
.
X Th

er

3 L
tin

vy (x) = − 3 (a + L) δ x + 2+3 x
a a a
on

.O
C

d) The stresses in zone ABCD of the fluid are computed by means of the con-
stitutive equation in Cartesian coordinates of Table 10.2. Using the expressions
for the pressure and velocity fields obtained in c) yields
⎡ ⎤
∂ vy
⎢ −p μ ∂ x 0 ⎥
⎢ ⎥
not ⎢ ∂v ⎥
σ ≡ ⎢ μ y −p 0 ⎥ where
⎢ ∂x


⎦ ∂ vy . x L 2

μ = μ δ −6 3 (a + L) + 3 2 + .
0 0 −p ∂x a a a

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
508 C HAPTER 10. F LUID M ECHANICS

C OMMENT
When the piston descends, the steady-state regime hypothesis is, in fact, not
completely rigorous since, at some point, the piston will reach the lowest point
of its trajectory and the flow will vary. To
. be able to apply this hypothesis, either
(m + h) must be a very large length or δ must be a very low velocity.

e) The stresses acting on the piston must be computed to obtain the resultant
forces and, then, the equilibrium of forces is applied to determine the expression
for F. These stresses are:

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
postive sign
f

ra
criterion
C d P cs
b
a
i
an an n

The stresses in the inferior surface of the piston are


y ha

  .
  6μ
le
σ ∗ = −σy  = p (y)  = k m +C1 =⇒ σ ∗ = 3 (a + L) hδ .
liv or ec

A y=m a
M

.A

In the lateral surfaces, due to symmetry, τ1∗ = τ2∗ and, therefore, only τ1∗ needs
m

to be computed,
d
uu

 . a 
e

 L 2
X Th

∗ ∗
er

τ1 = τ2 = −τxy  = −μ δ −6 3 (a + L) + 3 2 + =⇒
tin

x=a a a a
on

.O

. 
μδ L
τ1∗ = τ2∗ = 3 +4 .
C

a a
©

.
Imposing the equilibrium of forces (since δ is a constant velocity),
F = Lσ ∗ + h τ1∗ + h τ2∗ =⇒
 2 
. 2μh L L
F = ηδ with η = 3 2 +6 +4
a a a

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 509

E XERCISES

10.1 – Compute the horizontal and vertical components of the resultant of the
actions, per unit of length, exerted by the water on the gravity dam shown in the
figure.

rs
ee
s gin
t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
10.2 – The wall of a tank has a valve that rotates about point O as shown in the
i
an an n

figure. Compute the resultant force and moment, per unit of thickness, that the
y ha

fluid exerts on the valve. The weight of the valve can be neglected.
le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

10.3 – Determine the weight of the ballast W required at the bottom of the crate
shown in the figure, whose weight is W, such that it is maintained afloat in stable
equilibrium.

N OTE: The water has a density ρ and the weights are per unit of thickness.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
510 C HAPTER 10. F LUID M ECHANICS

10.4 – A container filled with water up to


a height H is placed on an inclined plane
with angle θ and dropped such that it slides
down this plane with a constant accelera-
tion value a. Determine the distribution of
pressures and the equation of the free sur-
face in terms of a, H, θ and the atmospheric
pressure pa .

rs
10.5 – A plate of indefinite size and thickness 2a separates two incompressible

ee
Newtonian fluids that move between two rigid boundaries of indefinite length

s gin
placed at a distance h from the plate, as shown in the figure. The plate and the
top boundary move at velocities v/2 and v, respectively. Determine:

t d le En
a) The pressure, velocity and stress fields in terms of the integration constants.

r
b) The integration constants, by applying the adequate boundary conditions.

ba
ge ro or
eS m
c) The forces per unit of surface F1 and F2 exerted on the plate and the top

ci
f

ra
boundary needed to produce the described motion.
C d P cs
b
a
d) The dissipated energy, per unit of time and of surface perpendicular to the
i
an an n

plane of the paper, due to viscous effects.


y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O

Additional hypotheses:
C

1) The pressures at points A and B are pA and pB , respectively.


2) Consider a steady-state regime.
3) Due to the indefinite character of the x-direction, the flow and its properties
can be considered to be invariable in this direction.

10.6 – A volume flow rate Q of an incompressible isotropic Newtonian fluid


flows in steady-state regime between the plate and the horizontal surface shown
in the figure. The plate is kept horizontal and immobile by means of a force
with horizontal and vertical components H and V, respectively, acting on an
appropriate point of said plate. Determine:

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 511

a) The pressure and velocity fields.


b) The value of the vertical component of the force and the distance d from the
origin to its application point, such that the plate does not rotate.
c) The value of the horizontal component of the force.

rs
ee
s gin
t d le En

r
ba
ge ro or
Additional hypotheses:

eS m
ci
f
1) The flow is assumed to be parallel to the x-y plane.

ra
C d P cs
2) Inertial forces can be neglected.
b
a
i
3) The volume flow rate, Q, and the components of the force, V and H, are
an an n

considered per unit of length in the z-direction.


y ha

4) The weight of the plate and the atmospheric pressure are negligible.
le
liv or ec
M

.A
m

10.7 – A cylindrical shell of indefinite length and internal radius R rotates in


d

steady-state regime at an angular velocity ω inside an infinite domain occupied


uu
e

by an incompressible Newtonian fluid with viscosity μ2 . A different incompress-


X Th

er
tin

ible Newtonian fluid, with viscosity μ1 , is contained inside the cylindrical shell.
Determine:
on

.O

a) The pressure and velocity fields


C

of the internal fluid.


©

b) The pressure and velocity fields


of the external fluid.
c) The moment that must be ap-
plied on the cylindrical shell to
maintain its velocity.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
512 C HAPTER 10. F LUID M ECHANICS

10.8 – A disc of radius R rotates with a constant angular velocity ω at a dis-


tance a from a horizontal surface. Between the disc and the surface is an incom-
pressible Newtonian fluid with viscosity μ. Determine:
a) The velocity field of the fluid in terms of the integration constants.
b) The value of the integration constants, by applying the appropriate boundary
conditions, and the complete expression of the velocity field.
c) The pressure field and the shear stress τzθ .
d) The value of the moment M that must be applied on the axis of the disc to
maintain the described motion.

rs
Additional hypotheses:

ee
1) The rotation of the disc is sufficiently

s gin
slow to neglect the inertial forces.
2) The effect of the lateral walls (fluid-

t d le En
wall friction effects) can be neglected.

r
3) The velocity field varies linearly with

ba
ge ro or
eS m
the distance to the inferior surface.

ci
f
4) Assume a steady-state regime.

ra
C d P cs
b
a
i
an an n

10.9 – The cross-section of a cylindrical piston ABA B that slides inside a con-
y ha

tainer filled with an incompressible Newtonian fluid with viscosity μ is shown


le
.
liv or ec

in the figure. The motion of the piston, at a velocity δ , causes the fluid to flow
through the pipe DED E .
M

.A

a) Determine the pressure and velocity fields of the fluid in zone DED E in
m

terms of the integration constants.


uu
e

b) Indicate and justify the boundary conditions that must be applied to deter-
X Th

er
tin

mine the value of the integration constants. Determine these constants and
the complete expressions of the pressure and velocity fields.
on

.O

c) Compute the stresses in zone DED E of the fluid.


C

d) Assuming that the stress normal to the surface BB in the fluid is constant
©

and equal to the pressure in points D and D , prove there exists


. a relation
between the force F applied .on the piston and its velocity δ , and that said
relation is of the form F = η δ . Determine the value of η.

Additional hypotheses:
1) The body forces of the fluid and weight
of the piston can be neglected.
2) Assume a steady-state regime.
3) The atmospheric pressure is negligible.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
CH.11. VARIATIONAL
PRINCIPLES
Multimedia Course on Continuum Mechanics
Overview
 Introduction Lecture 1
 Functionals
Lecture 2
 Gâteaux Derivative
 Extreme of a Functional Lecture 3
 Variational Principle
Lecture 4
 Variational Form of a Continuum Mechanics Problem
 Virtual Work Principle Lecture 5
 Virtual Work Principle
 Interpretation of the VWP Lecture 6
 VWP in Engineering Notation
 Minimum Potential Energy Principle
 Hypothesis Lecture 7
 Potential Energy Variational Principle

2
11.1. Introduction
Ch.11. Variational Principles

3
Computational Mechanics
 In computational mechanics problems are solved by cooperation
of mechanics, computers and numerical methods.
 This provides an additional approach to problem-solving, besides the
theoretical and experimental sciences.
 Includes disciplines such as solid mechanics, fluid dynamics,
thermodynamics, electromagnetics, and solid mechanics.

6
11.2. Functionals
Ch.11. Variational Principles

8
Definition of Functional
 Consider a function space X : R
b

∫ u ( x)dx
=X: {u ( x ) : Ω ⊂ R 3
→R m
} F (u )
b
a

 The elements of X are functions u ( x ) ∫ f [ x, u ( x), u′( x)] dx


a b
of an arbitrary tensor order, defined in X ∫ u′( x)dx
a subset Ω ⊂ R . u ( x)
3
a

u ( x ) : [ a, b ] → R

 A functional F ( u ) is a mapping of the function space X onto the


set of the real numbers , R : F ( u ) : X → R .
 It is a function that takes an element u ( x ) of the function space X as
its input argument and returns a scalar.

9
Definition of Gâteaux Derivative
 Consider :
 =
a function space X : {u( x ) : Ω ⊂ R 3 → R m }
 the functional F ( u ) : X → R
 a perturbation parameter ε ∈ R
 a perturbation direction η( x ) ∈ X

 The function u(x ) + ε η(x ) ∈ X is the perturbed function of u ( x ) in


the η ( x ) direction. t=0 t

u (x) Ω
P P’
Ω0 ε η(x)
u (x) + ε η(x)

10
Definition of Gâteaux Derivative
 The Gâteaux derivative of the functional F ( u ) in the η direction is:

( F ( u + εη ) )
d
( u; η) :
δF= P’
dε ε=0 F (u )
t=0
t

u (x) Ω
P P’
Ω0 ε η(x)
u (x) + ε η(x)

REMARK not
The perturbation direction is often denoted as η = δu .
Do not confuse δu(x) with the differential du(x) .
δu(x) is not necessarily small !!!

11
Example
Find the Gâteaux derivative of the functional t=0 t

F ( u )=: ∫ ϕ (u ) d Ω + ∫ φ (u ) d Γ u (x) Ω
Ω ∂Ω
P P’
Ω0 ε δu ( x )
u ( x ) + ε δu ( x )
Solution :

d d d
δF ( u; δ=
u) F ( u + εδu )= ∫ ϕ ( u + εδu ) d Ω + ∫ φ ( u + εδu ) d Γ=
dε d
ε=0 = ε Ω ε 0 =
d ε ∂Ω ε 0

 ∂ϕ ( u + εδu ) d ( u + εδu )   ∂φ ( u + εδu ) d ( u + εδu ) 


= ∫ ⋅ d Ω +  ∫ ⋅ d Γ
=Ω ∂u d ε ε 0 =
 ∂Ω ∂u d ε ε 0
= δu = δu

∂ϕ( u) ∂φ( u)
δF (=
u; δu ) ∫Ω ∂u ⋅ δu d Ω + ∫∂Ω ∂u ⋅ δu d Γ
12
Gâteaux Derivative with boundary
conditions
 Consider the function space V :
Γu  Γ σ =∅

=V: {u ( x ) u ( x ) : Ω → R m=
; u ( x ) x∈Γ u* ( x )
u
}

 By definition, when performing the Gâteaux derivative on V ,


( u + ε δu ) ∈ V .
 Then,
( u + ε δu ) x∈Γ u
= u* ( x ) u x∈Γ + ε δu x∈Γ = u*
u u
ε δu x∈Γ =0
u

=u *

 The direction perturbation must satisfy: δu x∈Γ =


0
u

13
Gâteaux Derivative in terms of
Functionals
 Consider the family of functionals
Γu  Γ σ =∅
=F (u ) ∫ φ (x, u(x), ∇u(x))d Ω

+ ∫ ϕ (x, u(x), ∇u(x))d Γ


Γσ

 The Gâteaux derivative of this family


of functionals can be written as,
∀δ u
δ F ( u; δ u )
= ∫ E(x, u(x), ∇u(x)) ⋅ δ u d Ω + ∫ T(x, u(x), ∇u(x)) ⋅ δ u d Γ δ u x∈Γ = 0
Ω Γσ
u

REMARK
The example showed that for F ( u )=: ∫ φ ( u ) d Ω + ∫ ϕ ( u ) d Γ , the
Ω ∂Ω

∂φ( u) ∂ϕ( u)
F (u)
Gâteaux derivative is δ= ∫Ω ∂u ⋅ δu d Ω + ∫∂Ω ∂u ⋅ δu d Γ .
14
Extrema of a Function
 A function f ( x ) : R → R has a local minimum (maximum) at x0

 Necessary condition:
df ( x) not
= f= ′ ( x0 ) 0
dx x = x0

Local minimum
 The same condition is necessary for the function to have extrema
(maximum, minimum or saddle point) at x0 .

 This concept can be can be extended to functionals.

15
Extreme of a Functional. Variational
principle
 A functional F ( u ) : V → R has a minimum at u ( x ) ∈ V

 Necessary condition for the functional to have extrema at u ( x ) :

δF ( u; δu ) = 0 ∀δ u | δ u x∈Γ = 0
u

 This can be re-written in integral form:

δ F ( u; δ u=
) ∫ E(u) ⋅ δ u d Ω + ∫ T(u) ⋅ δ u d Γ= 0 ∀δ u
Ω Γσ
δ u x∈Γ = 0
Variational Principle u

16
11.3.Variational Principle
Ch.11. Variational Principles

17
Variational Principle
 Variational Principle:

δ F ( u; δ u ) = ∫ E ⋅ δ u d Ω + ∫ T ⋅ δ u d Γ = 0 ∀δ u REMARK
Ω Γσ
δ u x∈Γ = 0 Note that δ u
is arbitrary.
u

 Fundamental Theorem of Variational Calculus:


The expression
∀δ u
∫Ω
E( x, u( x), ∇u( x)) ⋅ δ u d Ω + ∫ T( x, u( x), ∇u( x)) ⋅ δ u d Γ =
Γσ
0
δ u x∈Γ = 0
is satisfied if and only if u

E(x, u(x), ∇u(x=


)) 0 ∀x ∈ Ω Euler-Lagrange equations
T(x, u(x), ∇u(x=
)) 0 ∀x ∈ Γσ Natural boundary conditions

18
Example
Find the Euler-Lagrange equations and the natural and forced boundary
conditions of the functional
b

∫ φ  x, u ( x ) , u′ ( x ) dx with u ( x ) : [ a, b] → R ; u ( x ) x=a =


F (u ) =
a
u (a) =
p

19
Example - Solution
Find the Euler-Lagrange equations and the natural and forced boundary
conditions of the functional
b

(u )
F= ∫ φ  x, u ( x ) , u′ ( x ) dx with u ( x=
) x =a u=
(a) p
a
Solution :
First, the Gâteaux derivative must be obtained.
 The function u ( x ) is perturbed:

u ( x ) → u ( x ) + ε η ( x )  not
 ∀η ( x ) ≡ δ u ( x ) | η ( a ) = ηa =0
u ( x ) → u ( x ) + ε η ( x ) 
′ ′ ′
 This is replaced in the functional:
b
F (u=
+εη) ∫ φ  x, u ( x ) + ε η , u ′ ( x ) + ε η ′ dx
a

20
b
F ( u ) = ∫ φ  x, u ( x ) , u ′ ( x )  dx
Example - Solution u ( x=
a

(a) p
) x =a u=

 The Gâteaux derivative will be


 ∂φ ∂φ 
b
δ F ( u;η ) = ddε F ( u + ε η ) ε =0 =  η + η ′ dx

a 
∂u ∂u ′ 

Then, the expression obtained must be manipulated so that it resembles the


Variational Principle δ F ( u; δ u ) = ∫ E ⋅ δ u d Ω + ∫ T ⋅ δ u d Γ = 0:
Ω Γσ

 Integrating by parts the second term in the expression obtained:


∂φ ∂φ ∂φ ∂φ ∂φ ∂φ
b
b b d b d
∫a ∂u′
η ′dx = η − ∫ ( )η dx =
∂u′ a a dx ∂u′ ∂u′ b
ηb −
∂u′ a
ηa − ∫ ( )η dx
a dx ∂u′

ηa = 0
The Gâteaux derivative is re-written as:
b
 (u) φ  x, u ( x ) , u′ ( x )  dx ; u ( a )
∫= p
a

∂φ d ∂φ
b ∂φ
δ (u; η= ) δ (u; δ u ) = ∫ [ − ( )]δ udx + δ ub
a ∂u dx ∂u′ ∂u′ b
≡δ u
21
Example - Solution
Therefore, the Variational Principle takes the form
b  ∂φ d  ∂φ   ∂φ ∀δ u
δ (u; δ u ) = ∫  −   δ u dx + δ u =
0
a ∂u dx  ∂u′   ∂u ′ b δ ua = 0
b

If this is compared to δ F ( u; δ u ) = ∫ E ⋅ δ u d Ω + ∫ T ⋅ δ u d Γ = 0 , one obtains:


Ω Γσ

∂φ d  ∂φ 
E ( x, u , u ′ ) ≡ −  = 0 ∀x ∈ ( a, b ) Euler-Lagrange Equations

∂u dx  ∂u 

∂φ Natural (Newmann)
T ( x, u , u ′ ) ≡ =
0 boundary conditions
∂u ′ x =b

Essential (Dirichlet)
u ( x) x = a ≡ u (a) =
p
boundary conditions

22
Variational Form of a Continuum
Mechanics Problem
 Consider a continuum mechanics problem with local or strong
governing equations given by,
 Euler-Lagrange equations

E(x, u(x), ∇u(x))= 0 ∀x ∈ V

 with boundary conditions:


 Natural or Newmann
T(x, u(x), ∇u(x)) ≡  (∇u) ⋅ n − t* (x=
) 0 ∀x ∈ Γσ

 Forced (essential) or Dirichlet


u ( x ) u∗ ( x ) ∀x ∈ Γu
= REMARK
The Euler-Lagrange equations
are generally a set of PDEs.

23
Variational Form of a Continuum
Mechanics Problem
 The variational form of the continuum mechanics problem consists
in finding a field u ( x ) ∈ X where

=
V: {u ( x ) : V ⊂ R → R u=
3
( x ) u ( x ) on Γ }
m ∗
u

V= {δ u(x) : V ⊂ R → R δ u(x=
0
3
) 0 on Γ }
m
u

fulfilling:

∫ E(x, u(x), ∇u(x)) ⋅ δ u(x)dV + ∫


V Γσ
T( x, u( x), ∇u( x)) ⋅ δ u( x) d Γ =0 ∀δ u(x) ∈ V0

24
Variational Form of a Continuum
Mechanics Problem

REMARK 1
The local or strong governing equations of the continuum mechanics
are the Euler-Lagrange equation and natural boundary conditions.

REMARK 2
The fundamental theorem of variational calculus guarantees that the
solution given by the variational principle and the one given by the local
governing equations is the same solution.

25
11.4. Virtual Work Principle
Ch.11. Variational Principles

26
Governing Equations
 Continuum mechanics problem for a body:
 Cauchy equation
∂ 2u ( x, t )
∇ ⋅ σ ( x, t ) + ρ0b ( x, t ) =
ρ0 in V
∂t 2

(ε (∇u( x, t )))

 Boundary conditions

u ( x, t ) u∗ ( x, t ) on Γu
=
σ ( x, t ) ⋅ n (=
x, t ) t ∗ ( x, t ) on Γσ

 (ε (∇u),t)

27
Variational Principle
 The variational principle consists in finding a displacement field
, where= V : {u ( x, t ) : V ⊂ R 3 → R m u=
( x, t ) u∗ ( x, t ) on Γu }
such that the variational principle holds,
∂ 2u
δ W ( u; δ u=
) ∫ [∇ ⋅ σ + ρ (b − 2 )] ⋅ δ u dV + ∫ ( t∗ − σ ⋅ n) ⋅ δ u d Γ= 0 ∀δ u ∈ V0
∂t
V Γσ
=T
=E
where V0=: {δ u ( x ) : V ⊂ R3 → R m δ u ( x=) 0 on Γu }
 Note:
 is the space of admissible displacements.
 is the space of admissible virtual displacements (test functions).
 The (perturbations of the displacements ) δ u are termed virtual
displacements.

28
Virtual Work Principle (VWP)
 The first term in the variational principle
=a
∂ 2u
) ∫ [∇ ⋅ σ + ρ (b − 2 )] ⋅ δ u dV + ∫ ( t∗ − σ ⋅ n ) ⋅ δ u d Γ= 0 ∀δ u ∈ V0
δ W ( u; δ u=
∂t
V Γσ
=T
=E
Considering that ( ∇ ⋅ σ ) ⋅ δ u= ∇ ⋅ ( σ ⋅ δ u ) − σ : ∇ sδ u
and (applying the divergence theorem):

∫ ( ∇ ⋅ σ ) ⋅ δ=u ⋅ dV ∫ ( n ⋅ σ) ⋅ δ u dΓ − ∫  δ u dV
 σ : ∇ s

V Γσ V

 Then, the Virtual Work Principle reads:


δ W ( u; δ=
u) ∫ ρ ( b − a ) ⋅ δ u dV + ∫ t *
⋅ δ u d Γ − ∫ σ : ∇ s
δu =
dV 0 ∀δ u ∈ V0
V Γσ V

29
Virtual Work Principle (VWP)

REMARK 1
The Cauchy equation and the equilibrium of tractions at the boundary
are, respectively, the Euler-Lagrange equations and natural boundary
conditions associated to the Virtual Work Principle.

REMARK 2
The Virtual Work Principle can be viewed as the variational principle
associated to a functional W ( u ) , being the necessary condition to find
a minimum of this functional.

30
Interpretation of the VWP
 The VWP can be interpreted as:

δ W ( u=
;δ u ) ∫ ρ ( b

− a ) ⋅ δ u dV + ∫ t *
⋅ δ u d Γ − ∫ σ : ( ∇ s


) dV 0
δ u=
V Γσ V
b* δε
pseudo - virtual
body forces strains
Work by the
Work by the pseudo-body virtual strain.
forces and the contact forces. Internal virtual
External virtual work work
δW ext δ W int

δ W ( u; δ u ) = δ W ext − δ W int = 0 ∀δ u ∈ V0

31
VWP in Voigt’s Notation
 Engineering notation uses vectors instead of tensors:
σ x  δε x  δε x 
σ  δε  δε 
 y  y   y 
σ z  not δε
 z  δε z 
{σ} ∈ R ; {σ} =τ 
6
{δε } ∈ R ; {δε} =δγ  =2δε 
6
σ : δε ={σ} ⋅ {δε} ={δε} ⋅ {σ}
 xy   xy   xy 
τ  δγ  2δε 
 xz   xz   xz 
τ yz  δγ yz  2δε yz 

 The Virtual Work Principle becomes


 
δ W = ∫ {δε} ⋅ {σ} dV −  ∫ ρ ( b − a ) ⋅ δ u dV + ∫ t ⋅ δ u d Γ  = 0 ∀δ u ∈ V0
*
V 
V  Γσ 
Total virtual
work.
Internal virtual External virtual
work, δ W . work, δ W .
int ex t

32
11.5. Minimum Potential Energy Principle
Ch.11. Variational Principles

33
Hypothesis
 An explicit expression of the functional W in the VWP can only be
obtained under the following hypothesis:
1. Linear elastic material. The elastic potential is:
1 ∂uˆ (ε )
=
uˆ (ε ) ε:C:ε =σ = C:ε
2 ∂ε
2. Conservative volume forces. The potential for the quasi-static
case (a = 0) under gravitational ∂φ(u)
forces and constant density is: φ ( u ) =
− ρ b ⋅ u =−ρ b
∂u
3. Conservative surface forces. The potential is: ∂G(u)
G ( u ) =−t ∗ ⋅ u =−t ∗
∂u
 Then a functional, total potential energy, can be defined as
U ( u=
) ∫ uˆ (ε )dV + ∫ φ ( u ) dV + ∫ G ( u ) dS
V V Γσ

Elastic Potential energy of Potential energy of


s
= ε(∇ (u )) energy the body forces the surface forces
34
Potential Energy Variational Principle
 The variational form consists in finding a displacement field
u( x, t ) ∈ V , such that for any δ u =
δ u 0 in Γu the following
condition holds,
∂uˆ S ∂φ ( u ) ∂G ( u )
δ U ( u; δ u ) = ∫ : ∇ (δ u ) dV + ∫ ⋅ δ u dV + ∫ ⋅ δ u d Γ =0
∂ε ∂u ∂u
V
= δε V Γσ
*
= σ = −t
= −ρb

U ( u; δ u )
δ= ∫ σ : δε dV − ∫ ρ ( b − a ) ⋅ δ u dV − ∫ ⋅ δ u dΓ ∀δ u ∈ V0
*
t
V V Γσ

 This is equivalent to the VWP previously defined.


δ W ≡ δ U ( u; δ u )

35
Minimization of the Potential Energy
 The VWP is obtained as the variational principle associated with
this functional U , the potential energy.
deriving from a
 The potential energy is potential

1
=U(u) ∫V 2 ε (u ) : C : (u ) dV − ∫V ρ ( b − a (u ) ) ⋅ u dV − ∫Γ ⋅ u d Γ
t *

 This function has an extremum (which can be proven to be a minimum) for


the solution of the linear elastic problem.
 The solution provided by the VWP can be viewed in this case as
the solution which minimizes the total potential energy functional.

δ U(u; δ u) = 0 ∀δ u ∈ V0

36
Chapter 11
Variational Principles

rs
ee
s gin
11.1 Governing Equations

t d le En

r
Variational calculus is a mathematical tool that allows working with the so-

ba
ge ro or
eS m
called integral or weak form of the governing differential equations of a problem.

ci
f
Given a system of differential equations, which must be verified in local form

ra
C d P cs
(point by point) for a certain domain, the variational principles allow obtain-
b
a
i
ing an integral or weak formulation (global, in the domain), whose imposition,
an an n

nonetheless, guarantees that the aforementioned differential equations are sat-


y ha

isfied. Integral formulations are of particular interest when treating and solving
le
liv or ec

the problem by means of numerical methods.


M

.A

11.1.1 Functionals. Functional Derivatives


m

d
uu
e
X Th

er
tin

Definition 11.1. The functional F (u) is a mapping of the function


on

space X onto the set of real numbers R,


.O

  
C

F (u) : X → R where X := u (x)  u (x) : R3 ⊃ Ω → Rm .


©

In other words, the functional F (u) is a function that takes an ele-


ment u (x) (a scalar, vector or tensor function defined in a domain Ω
of R3 or, in general, Rn ) of a function space X as its input argument
and returns a real number.

With certain language misuse, one could say that the functional F (u) is a scalar
function whose arguments are functions u (x).

513
514 C HAPTER 11. VARIATIONAL P RINCIPLES

Example 11.1 – Consider an interval Ω ≡ [a, b] ∈ R and the space X


constituted by all the real functions with real variables in the interval
[a, b] (u (x) : [a, b] → R) with first derivatives u (x) that are integrable in this
interval. Examples of possible functionals are

b b
F (u) = u (x) dx , G (u) = u (x) dx
a a

b  

rs
and H (u) = F x, u (x) , u (x) dx .

ee
a

s gin
t d le En

r
ba
ge ro or
Definition 11.2. 
 Consider the (scalar, vector
 or tensor) function

eS m
space X := u (x)  u (x) : R3 ⊃ Ω → Rn on a domain Ω and a

ci
f

ra
functional F (•) : X → R.
C d P cs
b
a
Consider the two functions u, η ∈ X and the (perturbation) param-
i
an an n

eter ε ∈ R. Then, the function u + ε η ∈ X, can be interpreted as a


y ha

perturbed function of the function u in the direction η .


le
liv or ec

The Gateaux variation (or Gateaux derivative) of the functional


M

.A

F (u) in the direction η is defined as



m

de f d 
d

δ F (u; η ) = F (u + ε η )  .
uu
e

dε ε=0
X Th

er
tin
on

.O
C

Remark 11.1. The direction with respect to which the variation is


not
taken is often denoted as η = δ u. This notation will be used fre-
quently in the remainder of the chapter. Do not confuse δ u (x) with
the differential du (x) (in an infinitesimal calculus context) of a func-
tion u (x). However, obtaining the Gateaux variation of a functional
has in certain cases the same formalism as the ordinary differentia-
tion of functions and, thus, the risk of confusion (see Example 11.2).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Governing Equations 515

Example 11.2 – Obtain the Gateaux derivative of the functional


 
de f
F (u) = φ (u) dΩ + ϕ (u) dΓ .
Ω ∂Ω

Solution
  
d  d 
δ F (u; δ u) = 
F (u + εδ u)  = φ (u + εδ u) dΩ  +
dε ε=0 dε ε=0
Ω
 

rs
d 
+ ϕ (u + εδ u) dΓ  =

ee
dε ε=0
∂Ω

s gin
⎡ ⎤

t d le En
⎢ ∂ φ (u + εδ u) d (u + εδ u) ⎥
⎢ ⎥

r
=⎢ · dΩ ⎥ +

ba
ge ro or
⎣ ∂u dε ⎦

eS m
Ω 

ci
⎡ δu f ε =0 ⎤

ra
C d P cs
b
a
⎢  ∂ ϕ (u + εδ u) d (u + εδ u) ⎥
i
an an n

⎢ ⎥
+⎢ · dΓ ⎥ =⇒
y ha

⎣ ∂u dε ⎦

le
∂Ω
liv or ec

δu ε =0
M

.A

⎡ ⎤
   
m

∂ φ (u) ∂ ϕ (u)
δ⎣ ϕ (u) dΓ ⎦ =
d

φ (u) dΩ + · δ u dΩ + · δ u dΓ
uu

∂u ∂u
e
X Th

Ω ∂Ω Ω ∂Ω
er
tin

Note, in this case, the formal similarity of obtaining the Gateaux derivative
on

.O

of the functional with the differentiation of functions.


C

 
©

Consider now a domain Ω ⊂ R3 , its boundary ∂ Ω = Γu Γσ with Γu Γσ = 0/


(see Figure 11.1) and the space V of the functions u (x) defined on Ω and such
that they take a prescribed value u∗ (x) at the boundary Γu :
  
 

V := u (x) u (x) : Ω → R m
; u (x)  ∗
= u (x) (11.1)
x∈Γu

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
516 C HAPTER 11. VARIATIONAL P RINCIPLES


Γu Γσ = 0/

Figure 11.1: Definition of the domain Ω ⊂ R3 .

rs
ee
s gin
Remark 11.2. When computing the Gateaux derivative, a condition,
established in the definition itself, on the perturbation η ≡ δ u is that

t d le En
the perturbed function u + ε δ u must belong to the same function

r
space V (u + ε δ u ∈ V). In this case, if u + ε δ u ∈ V,

ba
ge ro or
eS m
   

ci
   
f

ra
(u + ε δ u)  = u∗ =⇒ u + ε δ u = u∗ =⇒ ε δ u = 0
C d P cs
b
a
x∈Γu x∈Γ x∈Γu x∈Γu
i
u
an an n

u∗ 
y ha


and the perturbation δ u must satisfy δ u =0.
le
liv or ec

x∈Γu
M

.A
m

Based on the family of functions (11.1), consider now the following family
uu
e

of functionals
X Th

er
tin

 
on

.O

F (u) = φ (x, u (x) , ∇u) dΩ + ϕ (x, u (x) , ∇u) dΓ ∀u ∈ V , (11.2)


C

Ω Γσ
©

where the functions φ and ϕ are regular enough to be integrable in the domains
Ω and Γσ , respectively. Assume, in addition, that, through adequate algebraic
operations, the Gateaux derivative of F (u) can be written as

 
δ F (u; δ u) = E (x, u (x) , ∇u) · δ u dΩ + T (x, u (x) , ∇u) · δ u dΓ
 . (11.3)
Ω Γσ 
∀δ u; δ u = 0
x∈Γσ

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Governing Equations 517

Example 11.3 – Obtain the Gateaux derivative, in the format given in (11.3),
of the functional

b   
 
F (u) = φ x, u (x) , u (x) dx with u (x)  = u (a) = p .
x=a
a

Solution
The given functional is a particular case of the functional in (11.2), reduced
to a single dimension with ϕ ≡ 0, Ω ≡ (a, b), Γu ≡ a and Γσ ≡ b.

rs
ee
Perturbing the function u (x) and replacing in the functional yields

s gin

u(x) → u(x) + ε η(x) 
 not
∀η (x) ≡ δ u (x)  η (a) = ηa = 0 =⇒

t d le En
  
u (x) → u (x) + ε η (x)

r
ba
ge ro or
eS m
b  

ci
F (u + ε η) = f
φ x, u (x) + ε η, u (x) + ε η  dx

ra
C d P cs
b
a
a
i
an an n

The Gateaux derivative is then


y ha

b  
le

liv or ec

d  ∂φ ∂φ 
δ F (u; η) = F (u + ε η)  = η +  η dx .
M

.A

dε ε=0 ∂u ∂u
a
m

On the other hand, the previous expression can be integrated by parts,


uu
e
X Th

b  x=b b    
er
tin

∂φ  ∂φ d ∂φ
η dx = η − η dx =
∂ u ∂ u x=a dx ∂ u
on

.O

a a
 
 b    
∂ φ 
C

∂φ  d ∂φ
=   ηb −  ηa − η dx =
©

∂u  ∂ u   dx ∂ u
x=b x=a
 =0 a
 b    
∂φ  d ∂φ
=   ηb − η dx ,
∂u  dx ∂ u
x=b a

producing the expression



b   
  ∂φ d ∂φ ∂ φ 
δ F u; δ u = − δ u dx +   δ ub ,
 ∂ u dx ∂ u ∂u 
η a x=b

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
518 C HAPTER 11. VARIATIONAL P RINCIPLES

which is a particular case of (11.3) with

 
∂φ d ∂φ
E (x, u, u ) ≡ − ∀x ∈ (a, b)
∂ u dx ∂ u
 .
∂ φ 

T (x, u, u ) ≡  
∂u 
x=b

rs
11.1.2 Extrema of the Functionals. Variational Principles.

ee
Euler-Lagrange Equations

s gin
Consider a real function of a real variable f (x) : R → R. This function is said to

t d le En
have a minimum at x = x0 when

r
ba
ge ro or
f (x0 ) ≤ f (x) ∀x ∈ R . (11.4)

eS m
ci
f

ra
The necessary condition for f to have an extrema (maximum, minimum or sad-
C d P cs
b
a
dle point) at x = x0 is known to be
i
an an n


d f (x) 
y ha

not 
 = f (x0 ) = 0 .
le
(11.5)
dx 
liv or ec

x=x0
M

.A

This concept can be extended to the functionals in a function space. Given a


m

functional F (u) : V → R, this functional is said to have a minimum at u (x)


d
uu

when
e

F (u) ≤ F (v) ∀v ∈ V ,
X Th

(11.6)
er
tin

and a necessary condition for the functional to have an extreme (maximum,


on

.O

minimum or saddle point) at u (x) is that the derivative δ F (u; δ u) be null in


every direction δ u,
C

 
δ F (u; δ u) = 0 ∀δ u  δ u =0 . (11.7)
x∈Γu

Expressing (11.7) in the same format as (11.3) results in

Variational principle
 
δ F (u; δ u) = E · δ u dΩ + T · δ u dΓ = 0  (11.8)
 
Ω Γσ ∀δ u  δ u =0
x∈Γu

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Governing Equations 519

Theorem 11.1. Fundamental Theorem of Variational Calculus:


Given E (x) : Ω → Rm and T (x) : Γσ → Rm that satisfy
   

E (x) · δ u dΩ + T (x) · δ u dΓ = 0 ∀δ u  δ u =0
x∈Γu
Ω Γσ

E (x) = 0 ∀x ∈ Ω
⇐⇒
T (x) = 0 ∀x ∈ Γσ

rs
ee
s gin
Proof (indicative1 )

t d le En
Consider the following choice for δ u (x).

r
⎨ E (x) ∀x ∈ Ω

ba
ge ro or
eS m
ci
δ u (x) = 0 ∀x ∈ Γu
⎩ f

ra
C d P cs
T (x) ∀x ∈ Γσ
b
a
i
an an n

Replacing in the theorem results in


y ha

 
le
liv or ec

E (x) · E (x) dΩ + T (x) · T (x) dΓ = 0 ⇐⇒ E (x) = T (x) = 0 .


 Γσ 
M

.A

Ω
≥0 ≥0
m

Q.E.D.
uu
e
X Th

Equation (11.8) is denoted as variational principle2 and, since δ u is arbitrary,


er
tin

in accordance with Theorem 11.1 it is completely equivalent to


on

.O
C

Euler-Lagrange equations
(11.9)
©

E (x, u (x) , ∇u (x)) = 0 ∀x ∈ Ω

Natural boundary conditions


T (x, u (x) , ∇u (x)) = 0 ∀x ∈ Γσ (11.10)

1 This proof is not rigorous and is provided solely as an intuitive indication of the line of
reasoning followed by the theorem’s proof.
2 Strictly speaking, (11.8) is a variational equation or the weak form of a differential prob-
lem.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
520 C HAPTER 11. VARIATIONAL P RINCIPLES

Remark 11.3. Equations (11.9),


E (x, u (x) , ∇u (x)) = 0 ∀x ∈ Ω ,

are, in general, a set of partial differential equations (PDEs) known


as Euler-Lagrange equations of the variational principle (11.8).
Equations (11.10),

T (x, u (x) , ∇u (x)) = 0 ∀x ∈ Γσ ,

rs
ee
constitute a set of boundary conditions on these differential equa-

s gin
tions denoted as natural or Neumann boundary conditions. Together
with the conditions (11.1),

t d le En
u (x) = u∗ (x) ∀x ∈ Γu ,

r
ba
ge ro or
eS m
named essential or Dirichlet boundary conditions, they define a sys-

ci
f
tem whose solution u (x) is an extreme of the functional F.

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec

Example 11.4 – Obtain the Euler-Lagrange equations and the corresponding


M

natural and essential boundary conditions of the functional in Example 11.3,


.A
m

b   

d

F (u) = φ x, u (x) , u (x) dx u (x)  = u (a) = p .


uu

with
e

x=a
X Th

er

a
tin
on

.O

Solution
C

From the result of Example 11.3,


©


b   
  ∂φ d ∂φ ∂ φ 
δ F u; δ u = − δ u dx +   δ ub ,
 ∂ u dx ∂ u ∂u 
η a x=b

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Virtual Work Principle (Theorem) 521

one directly obtains:

Euler-Lagrange equations :  
 ∂φ d ∂φ
E (x, u, u ) ≡ − =0 ∀x ∈ (a, b)
∂ u dx ∂ u

Natural boundary conditions : 


∂ φ 

T (x, u, u ) ≡   =0
∂u 

rs
x=b

ee
Essential boundary conditions
 :

s gin

u (x)  = u (a) = p
x=a

t d le En

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
11.2 Virtual Work Principle (Theorem)
i
an an n
y ha

Consider a material volume of the continuous medium Vt , occupying at time t


the volume in space V , subjected to the body forces b (x,t) and the surface
le
liv or ec

forces t∗ (x,t) on the boundary Γσ (see Figure 11.2). Consider also the func-
M

.A

tional spaceV of all the admissible displacements, which satisfy the boundary
condition ux∈Γu = u∗ .
m

d
uu
e
X Th

Space of admissible displacements


er
tin

  
  (11.11)
V := ut (x) : V → R3  ut (x)  = ut∗ (x)
on

.O

x∈Γu
C

Two of the equations governing the behavior of the medium are


Cauchy’s equation: ∇ · σ (u) + ρ (b − a (u)) = 0 ∀x ∈ V , (11.12)
Equilibrium condition
σ (u) · n − t∗ = 0 ∀x ∈ Γσ , (11.13)
at the boundary Γσ :
where the implicit dependency of the stresses on the displacements (through the
strains and the constitutive equation σ (u) = σ (εε (u)) ) and of the accelerations
on the displacements (through equation a (x,t) = ∂ 2 u (x,t)/∂ t 2 ) has been taken
into account.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
522 C HAPTER 11. VARIATIONAL P RINCIPLES

rs
ee
s gin
Figure 11.2: Definition of the material volume Vt .

t d le En
Consider now the variational principle

r
ba
ge ro or
eS m
  

ci
f
∇ · σ (u) + ρ (b − a (u)) · δ u dV +

ra
δ W (u; δ u) =
C d P cs
b
a
V 
i
E
an an n

    (11.14)
y ha

 
+ t∗ − σ (u) · n · δ u dΓ = 0 ; ∀δ u (x)  δ u =0,
le
liv or ec

x∈Γu
Γσ 
M

.A

T
m

where the displacement perturbations δ u are denoted as virtual displacements.


d
uu
e

 
X Th


er

Virtual displacements: δ u : V → R3  δ u
tin

=0 (11.15)
x∈Γu
on

.O

In view of (11.8) and (11.9), the Euler-Lagrange equations of the variational


C

principle (11.14) and their natural boundary conditions are


©

Euler-Lagrange
E ≡ ∇ · σ + ρ (b − a) = 0 ∀x ∈ Ω ,
equations:
(11.16)
Natural boundary T ≡ t∗ − σ · n = 0 ∀x ∈ Γσ ,
conditions:

that is, Cauchy’s equation (11.12) and the equilibrium condition at the bound-
ary (11.13).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Virtual Work Principle (Theorem) 523

The variational principle (11.14) can be rewritten in a totally equivalent form


as follows. Consider the term




⎪ (∇ · σ ) · δ u = ∇ · (σ σ · δ u) − σ : (∇ ⊗ δ u) = ∇ · (σ σ · δ u) − σ : (δ u ⊗ ∇)



∂ σi j ∂ (σi j δ u j ) ∂ (δ u j ) ∂ (σi j δ u j ) ∂ (δ u j )

⎪ δuj = − σi j = − σ ji

⎪ ∂ xi ∂ xi ∂ xi ∂ xi ∂ xi


⎩ i, j ∈ {1, 2, 3}
(11.17)

rs
and the splitting of δ u ⊗ ∇ into its symmetrical part, ∇s δ u, and its skew-

ee
symmetric part ∇a δ u,

s gin
δ u ⊗ ∇ = ∇ s δ u + ∇a δ u
with

t d le En
de f 1 de f 1
∇s δ u = (δ u ⊗ ∇ + ∇ ⊗ δ u) and ∇a δ u = (δ u ⊗ ∇ − ∇ ⊗ δ u) .

r
2 2

ba
ge ro or
eS m
(11.18)

ci
Introducing (11.18) in (11.17)3 produces
f

ra
C d P cs
b
a
i
σ · δ u) − σ : (δ u ⊗ ∇) =
(∇ · σ ) · δ u = ∇ · (σ
an an n
y ha

σ · δ u) − σ : ∇s δ u − σ : ∇a δ u
= ∇ · (σ =⇒ (11.19)
le

liv or ec

=0
M

.A
m

σ · δ u) − σ : ∇s δ u
(∇ · σ ) · δ u = ∇ · (σ . (11.20)
d
uu
e
X Th

Integrating now (11.20) over the domain V and applying the Divergence Theo-
er
tin

rem yields
on

  
.O

(∇ · σ ) · δ u dV = σ · δ u) dV −
∇ · (σ σ : ∇s δ u dV =
C

V  V  V

= σ · δ u) dΓ −
n · (σ σ : ∇s δ u dV =

∂V =Γu Γσ
 V 
= (n · σ ) · δ u dΓ + (n · σ ) · δ u dΓ − σ : ∇s δ u dV =⇒

Γu =0 Γσ V
(11.21)

3 The tensor σ is symmetrical and the tensor ∇a δ u is skew-symmetric. Consequently, their


product is null, σ : ∇a δ u = 0.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
524 C HAPTER 11. VARIATIONAL P RINCIPLES

  
(∇ · σ ) · δ u dV = (n · σ ) · δ u dΓ − σ : ∇s δ u dV , (11.22)
V Γσ V


where the condition δ ux∈Γu = 0 (see (11.15)) has been taken into account. Fi-
nally, introducing (11.20) in the original form of the variational principle (11.14)
results in
     
δ W (u; δ u) = ∇ · σ + ρ (b − a) · δ u dV + t∗ − (σ
σ · n) · δ u dΓ =

rs
 V   Γσ 

ee

= (∇ · σ ) · δ u dV + ρ (b − a) · δ u dV + t · δ u dΓ − σ · n) · δ u dΓ =

s gin
V  V Γσ Γσ

t d le En
=− σ : ∇ δ u dV +
s
ρ (b − a) · δ u dV + t · δ u dΓ = 0 =⇒

r
V V Γσ

ba
ge ro or
eS m
(11.23)

ci
f

ra
Virtual Work Principle
C d P cs
 
b
a
δ W (u; δ u) = ρ (b − a) · δ u dV + t∗ · δ u dΓ
i
an an n

(11.24)
y ha

V
 Γσ 
 
le
∀δ u (x) δ u = 0

liv or ec

− σ : ∇s δ u dV = 0
M

x∈Γu
.A

V
m

Expression (11.24), which is completely equivalent to the original variational


d
uu

principle and maintains the same Euler-Lagrange equations and boundary con-
e
X Th

ditions (11.16), is known as the Virtual Work Principle (or Theorem) (VWP).
er
tin
on

.O

Remark 11.4. The VWP is a variational principle frequently applied


C

in solid mechanics that can be interpreted as the search of an extrema


©

of a functional of a displacement field W (u), not necessarily known


in its explicit form, whose variation (Gateaux derivative) δ W (u; δ u)
is known and is given by (11.14). Since the Euler-Lagrange equa-
tions of the VWP are the Cauchy’s equation (11.12) and the equilib-
rium condition at the boundary (11.13), its imposition is completely
equivalent (yet, more convenient when solving the problem through
numerical methods) to the imposition in local form of the aforemen-
tioned equations and receives the name of weak form of these equa-
tions.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Virtual Work Principle (Theorem) 525

Remark 11.5. The constitutive equation does not intervene in the


VWP formulation and the type of kinematics considered (finite or
infinitesimal strains) is not distinguished either. Thus, the applica-
tion of the VWP is not restricted by the type of constitutive equation
chosen (elastic, elastoplastic, fluid, etc.) nor by the kinematics (finite
or infinitesimal strains) considered.

rs
ee
11.2.1 Interpretation of the Virtual Work Principle

s gin
Consider the continuous medium in the present configuration Vt at time t sub-
jected to the fictitious body forces b∗ (x,t) = b (x,t) − a (x,t) and the real sur-

t d le En
face forces t∗ (x,t) (see Figure 11.3), and suffering the real stresses σ (x,t).

r
ba
ge ro or
Consider, in addition, the virtual (fictitious) configuration Vt+δt corresponding

eS m
to time t + δt, separated from the real configuration by a virtual displacement

ci
f

ra
field (11.15)
C d P cs
b
a

i

an an n

 
Virtual displacements: δ u δ u =0 . (11.25)
y ha

x∈Γu
le
liv or ec
M

.A
m

t − present configuration
uu
e

t + δt − virtual configuration
X Th

er
tin
on

.O
C

Figure 11.3: Continuous medium subjected to fictitious body forces and real surface
forces.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
526 C HAPTER 11. VARIATIONAL P RINCIPLES

Under infinitesimal strain kinematics, the virtual strains associated with the
virtual displacements (11.25) are

Virtual strains: δ ε = ∇s δ u (11.26)

and, assuming that the stresses σ (x,t) remain constant along the time inter-
val [t, t + δt], the virtual strain work (internal virtual work) performed by the
medium during this interval is

 

rs
Internal δ Wint = σ : δ ε dV = σ : ∇s δ u dV .

ee
(11.27)
virtual work:

s gin
V V

Likewise, assuming that both the pseudo-body forces b∗ (x,t) and the surface

t d le En
forces t∗ (x,t) remain constant during the virtual strain process in the inter-

r
val [t, t + δt], the work performed by these forces (external virtual work) results

ba
ge ro or
eS m
in

ci
f

ra
C d P cs
 
b
a
External δ Wext = ρ (b − a) · δ u dV = t∗ · δ u dV
i
an an n

virtual work:  (11.28)


y ha

V
b∗ Γσ
le
liv or ec

and, comparing the VWP (11.24) with expressions (11.27) and (11.28), the
M

.A

VWP can be interpreted as follows.


m

d
uu
e

Virtual Work Principle


  
X Th

er
tin

δW = σ : δ ε dV − ρ (b − a) · δ u dV + t∗ · δ u dΓ = 0

on

.O

V V Γσ
Total
virtual  
C

work Internal 
 virtual External virtual
work δ Wint work (δ Wext )
©

(11.29)
=⇒ δ W = δ Wint − δ Wext = 0
for any kinematically admissible
change inthe virtual configuration



δ u =0
x=Γu

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Virtual Work Principle (Theorem) 527

11.2.2 Virtual Work Principle in terms of the Stress and Strain Vectors
The vectors of stress {σ σ } and virtual strain {δ ε } can be extracted from the
symmetrical tensors of stress, σ , and virtual strain, δ ε = ∇s δ u, in (11.29) as
follows.
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
σx δ εx δ εx
⎢σ ⎥ ⎢ δε ⎥ ⎢ δε ⎥
⎢ y⎥ ⎢ y⎥ ⎢ y ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢ σ ⎥ ⎢ δ ε ⎥ ⎢ δ ε z ⎥
σ} ≡ ⎢ z ⎥ {δ ε } ∈ R6 ; {δ ε } ≡ ⎢ z ⎥ = ⎢
not not
σ } ∈ R6 ; {σ
{σ ⎥
⎢ τxy ⎥ ⎢ δ γxy ⎥ ⎢ 2δ εxy ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣ τxz ⎦ ⎣ δ γxz ⎦ ⎣ 2δ εxz ⎦

rs
τyz δ γyz 2δ εyz

ee
(11.30)

s gin
They satisfy the equality
"

t d le En
σ : δ ε = {σ σ } · {δ ε } = {δ ε } · {σ
σ} ,

r
(11.31)

ba
ge ro or
σi j δ εi j = σm δ εm = δ εm σm i, j ∈ {1, 2, 3} , m ∈ {1, ..6} .

eS m
ci
f

ra
C d P cs
Replacing (11.31) in the VWP (11.29) results in
b
a
i
an an n
y ha

Virtual Work Principle


  
le
liv or ec

σ } dV − ρ (b − a) · δ u dV + t∗ · δ u dΓ = 0
δ W = {δ ε } · {σ

M

.A

Total V V Γσ
virtual  
m

work Internal 
 virtual External virtual
d

work δ Wint work (δ Wext )


uu

(11.32)
e
X Th

=⇒ δ W = δ Wint − δ Wext
er
tin

for any kinematically admissible


on

.O

change inthe virtual configuration





C

δ u =0
©

x=Γu

which constitutes the VWP form most commonly used in engineering.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
528 C HAPTER 11. VARIATIONAL P RINCIPLES

11.3 Potential Energy. Minimum Potential Energy


Principle
The functional W, in terms of which the variational principle (11.24) is estab-
lished, can be explicitly formulated only under certain circumstances. One such
case requires the following conditions:
1) Linear elastic problem
The constitutive equation can be written in terms of the elastic potential û (εε ) as
follows4 .

rs
ee
Elastic 1 1
û (εε ) = ε :C :ε = σ :ε
 2

s gin
potential: 2
σ (11.33)
∂ û (εε )

t d le En
=C :ε =σ
∂ε

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
2) The body forces ρb∗ (x,t) are conservative
b
a
i
That is, these body forces derive from a potential φ (u) and, thus,
an an n
y ha

le
∂ φ (u)
liv or ec

= −ρb∗ = −ρ (b − a) . (11.34)
∂u
M

.A
m

d
uu
e

Remark 11.6. A typical case of conservative body forces is obtained


X Th

er

for the quasi-static case (a = 0) under gravitational forces and con-


tin

stant density,
on

.O

b (x,t) ≡ [ 0, 0, −g ]T = const.
not
and ρ (x,t) = const.
C

In this case, the potential of the body forces is


∂ φ (u)
φ (u) = −ρb · u =⇒ = −ρb .
∂u

4 The restriction to the linear elastic problem can be made less strict and be extended to the
case of hyperelastic materials in a finite strain regime.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Potential Energy. Minimum Potential Energy Principle 529

3) The surface forces t∗ (x,t) are conservative


Therefore, they derive from a potential G (u) such that

∂ G (u)
t∗ = − . (11.35)
∂u

Remark 11.7. A typical case of conservative surface forces occurs


when the traction vector t∗ (x,t) is independent of the displacements,

rs
ee
∂ t∗
=0.

s gin
∂u
In this case, the potential of the surface forces is

t d le En
∂ G (u)

r
G (u) = −t∗ · u = −t∗ .

ba
=⇒

ge ro or
eS m
∂u

ci
f

ra
C d P cs
b
a
i
an an n
y ha

Under the above circumstances, the following functional, named total poten-
le
liv or ec

tial energy, can be defined.


M

.A

Potential energy
m

  
d
uu

U (u) = û (εε (u)) dV + φ (u) dV + G (u) dΓ


e

 (11.36)
X Th

er

Γσ
tin

V V
Total
potential   
energy Elastic Potential energy Potential energy of
on

.O

energy of the body forces the surface forces


C

whose Gateaux variation is


  
∂ û ∂ φ (u) ∂ G (u)
δ U (u; δ u) = : ∇S (δ u) dV + · δ u dV + · δ u dΓ =
∂ ε 
 ∂ u  ∂ u 
Γσ
V
σ δε V
−ρ (b − a) −t∗
    
= σ : δ ε dV − ρ (b − a) · δ u dV − t∗ · δ u dΓ ; ∀δ u  δ u =0,
x∈Γu
V V Γσ
(11.37)
where (11.33) to (11.35) have been taken into account.

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
530 C HAPTER 11. VARIATIONAL P RINCIPLES

Comparing (11.37) with the VWP (11.29) leads to


  
δ W ≡ δ U (u; δ u) = σ : δ ε dV − ρ (b − a) · δ u dV − t∗ · δ u dΓ = 0
V V
  Γσ
 
∀δ u δ u =0.
x∈Γu
(11.38)

Definition 11.3. Minimum Potential Energy Principle:

rs
The variational principle (11.38), which is still the weak form of

ee
Cauchy’s equation (11.12) and the equilibrium condition at the
boundary (11.13), is now the Gateaux variation of the potential

s gin
energy functional U (u) in (11.36). Consequently, this functional,

t d le En
which for the case of constant body and surface forces takes the form
  

r
1

ba
ε : C : ε dV − ρ (b − a) · u dV − t∗ · u dΓ ,

ge ro or
U (u) =

eS m
2

ci

V V
f Γσ

ra
û (εε )
C d P cs
b
a
i
an an n

presents an extreme (which can be proven to be a minimum5 ) for the


y ha

solution to the linear elastic problem.


le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

5 The condition of minimum of an extreme is proven by means of the thermodynamic re-


quirement that C be positive-definite (see Chapter 6).

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 531

P ROBLEMS

Problem 11.1 – From the expression of the Virtual Work Principle,


   
σ : δ ε dV0 = ρb · δ u dV0 + t · δ u dΓ ∀ δu  δu = 0 in Γu ,
V0 V0 Γσ

rs
prove the Minimum Potential Energy Principle for a linear elastic material

ee
under infinitesimal strain regime.

s gin
t d le En

r
Solution

ba
ge ro or
eS m
ci
A linear elastic material is a particular type of hyperelastic material and, thus,
f

ra
there must exist an elastic potential of the type
C d P cs
b
a
 ∂W
i

an an n

∃ W (εε )  = σi j =⇒ δ W = σi j δ εi j = σ : δ ε .
y ha

∂ εi j
le
liv or ec

In addition, if the external forces are conservative, the following is satisfied:


M

.A

 ∂ G (u)

∃ G (u)  t∗ = − =⇒ δ G = −t∗ · δ u
m

∂u
d
uu


e

 ∂ Φ (u)
X Th

er

∃ Φ (u)  ρb = − =⇒ δ Φ = −ρb · δ u
tin

∂u
on

.O

Now, the given expression of the Virtual Work Principle can be rewritten as
   
C

δ W dV0 + δ Φ dV0 + δ G dΓ = 0 ∀ δu  δu = 0 in Γu ,
©

V0 V0 Γσ
⎛ ⎞
   
δ⎝ W dV0 + Φ dV0 + G dΓ ⎠ = 0 ∀ δu  δu = 0 in Γu .
V0 V0 Γσ

Defining the total potential energy as


  
U (u) = W dV0 + Φ dV0 + G dΓ
V0 V0 Γσ

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
532 C HAPTER 11. VARIATIONAL P RINCIPLES

leads to 
δU = 0 ∀ δu  δu = 0 in Γu ,
which is the same as stating that U has an extreme at u. To prove that this extreme
is a minimum, consider
1 ∂ 2W
W (εε ) = ε : C : ε where Ci jkl = .
2 ∂ εi j ∂ εkl

Then, the expressions for U (u) and U (u + δ u) are computed as


  

rs
1
U (u) = ε (u) : C : ε (u) dV0 − ρb · (u) dV0 − t∗ · u dΓ and

ee
2
V0 V0 Γσ

s gin
 
1
U (u + δ u) = ε (u + δ u) : C : ε (u + δ u) dV0 − ρb · (u + δ u) dV0

t d le En
2


r
V0 V0

ba
ge ro or
t∗ · (u + δ u) dΓ .

eS m

ci
f

ra
Γσ
C d P cs
b
a
Taking into account
i
an an n

ε (u + δ u) = ε (u) + ε (δ u)
y ha

results in the following expression for the subtraction U (u + δ u) − U (u):


le
liv or ec

 
M

.A

1 1
U (u + δ u) − U (u) = ε (u) : C : ε (δ u) dV0 + ε (δ u) : C : ε (u) dV0
m

2 2
d

V0 V0
uu

  
e

1
t∗ · δ u dΓ
X Th

+ ε (δ u) : C : ε (δ u) dV0 − ρb · δ u dV0 −
er
tin

2
V0 V0 Γσ
on

.O

Introducing
C

ε (u) : C : ε (δ u) = ε (δ u) : C : ε (u) = σ : δ ε
reduces the subtraction to
 
1
U (u + δ u) − U (u) = σ : δ ε dV0 + ε (δ u) : C : ε (δ u) dV0
2
V0  V0 
− ρb · δ u dV0 − t∗ · δ u dΓ .
V0 Γσ

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
Problems and Exercises 533

Now, considering the previous expression and


  
δU = σ : δ ε dV0 − ρb · δ u dV0 − t∗ · δ u dΓ
V0 V0 Γσ

yields 
1
U (u + δ u) − U (u) = ε (δ u) : C : ε (δ u) dV0 .
2
V0

Finally, since the tensor Ci jkl = ∂ 2 W/ (∂ εi j ∂ εkl ) is positive-definite,

rs
ee
U (u + δ u) − U (u) ≥ 0

s gin
t d le En
and, thus, the potential energy is seen to have a minimum in the equilibrium
state.

r
ba
ge ro or
eS m
ci
f

ra
C d P cs
b
a
i
an an n
y ha

le
liv or ec
M

.A
m

d
uu
e
X Th

er
tin
on

.O
C

X. Oliver and C. Agelet de Saracibar Continuum Mechanics for Engineers.Theory and Problems
doi:10.13140/RG.2.2.25821.20961
TENSOR ALGEBRA
Multimedia Course on Continuum Mechanics
Overview
 Introduction to tensors Lecture 1

 Indicial or (Index) notation Lecture 2

 Vector Operations Lecture 3 Lecture 4

 Tensor Operations Lecture 5

 Differential Operators Lecture 6

 Integral Theorems
Lecture 7
 References

2
Introduction

SCALAR ρ , θ , ...
v
VECTOR v, f , ...

MATRIX σ , ε, ...

? C, ...

4
Concept of Tensor
 A TENSOR is an algebraic entity with various components
which generalizes the concepts of scalar, vector and matrix.

 Many physical quantities are mathematically represented as tensors.

 Tensors are independent of any reference system but, by need, are


commonly represented in one by means of their “component matrices”.

 The components of a tensor will depend on the reference system


chosen and will vary with it.

5
Order of a Tensor
 The order of a tensor is given by the number of indexes
needed to specify without ambiguity a component of a tensor.

a  Scalar: zero dimension α = 3.14  1.2 


 
a , a  Vector: 1 dimension vi =  0.3 
 0.8 
A, A  2nd order: 2 dimensions    0.1 0 1.3 
 
Eij =  0 2.4 0.5 
A , A  3rd order: 3 dimensions  1.3 0.5 5.8 
 
A , A  4th order …

6
Cartesian Coordinate System
 Given an orthonormal basis formed by three mutually
perpendicular unit vectors:
eˆ1 ⊥ eˆ 2 , eˆ 2 ⊥ eˆ 3 , eˆ 3 ⊥ eˆ1
Where:

=eˆ1 1=
, eˆ 2 1=
, eˆ 3 1

 Note that

1 if i = j 
=
eˆ i ⋅ eˆ j  =  δ ij
0 if i ≠ j 

7
Indicial or (Index) Notation
Tensor Algebra

10
Tensor Bases – VECTOR
 A vector v can be written as a unique linear combination of the
three vector basis eˆ i for i ∈ {1, 2,3} .
v
v = v1eˆ1 + v 2eˆ 2 + v3eˆ 3
v3
 In matrix notation:
 v1 
[ v ] =  v2  v1
 v3  v2
 In index notation:

v = ∑ vi eˆ i tensor as a physical entity


i
[ v ]i = vi component i of the tensor in the
given basis i ∈ {1, 2,3}

11
Tensor Bases – 2nd ORDER TENSOR
 A 2nd order tensor A can be written as a unique linear combination
of the nine dyads eˆ i ⊗ eˆ j ≡ eˆ i eˆ j for i, j ∈ {1, 2,3} .

= A11 ( eˆ1 ⊗ eˆ1 ) + A12 ( eˆ1 ⊗ eˆ 2 ) + A13 ( eˆ1 ⊗ eˆ 3 ) +


A
+ A21 ( eˆ 2 ⊗ eˆ1 ) + A22 ( eˆ 2 ⊗ eˆ 2 ) + A23 ( eˆ 2 ⊗ eˆ 3 ) +
+ A31 ( eˆ 3 ⊗ eˆ1 ) + A32 ( eˆ 3 ⊗ eˆ 2 ) + A33 ( eˆ 3 ⊗ eˆ 3 )

Alternatively, this could have been written as:

A = A11 eˆ1eˆ1 + A12 eˆ1eˆ 2 + A13 eˆ1eˆ 3 +


+ A21 eˆ 2eˆ1 + A22 eˆ 2eˆ 2 + A23 eˆ 2eˆ 3 +
+ A31 eˆ 3eˆ1 + A32 eˆ 3eˆ 2 + A33 eˆ 3eˆ 3

12
Tensor Bases – 2nd ORDER TENSOR
= A11 ( eˆ1 ⊗ eˆ1 ) + A12 ( eˆ1 ⊗ eˆ 2 ) + A13 ( eˆ1 ⊗ eˆ 3 ) +
A
+ A21 ( eˆ 2 ⊗ eˆ1 ) + A22 ( eˆ 2 ⊗ eˆ 2 ) + A23 ( eˆ 2 ⊗ eˆ 3 ) +
+ A31 ( eˆ 3 ⊗ eˆ1 ) + A32 ( eˆ 3 ⊗ eˆ 2 ) + A33 ( eˆ 3 ⊗ eˆ 3 )
 In matrix notation:
 A11 A12 A13 
[ A ] =  A21 A22 A23 
 A31 A32 A33 
 In index notation:
=A ∑ Aij ( eˆ i ⊗ eˆ j )
ij
tensor as a
physical entity

[ A ]ij = Aij component ij of the tensor


in the given basis i, j ∈ {1, 2,3}

13
Tensor Bases – 3rd ORDER TENSOR
 A 3rd order tensor A can be written as a unique linear combination
of the 27 tryads eˆ i ⊗ eˆ j ⊗ eˆ k ≡ eˆ i eˆ j eˆ k for i, j , k ∈ {1, 2,3}.
A A 111 ( eˆ1 ⊗ eˆ1 ⊗ eˆ1 ) + A 121 ( eˆ1 ⊗ eˆ 2 ⊗ eˆ1 ) + A 131 ( eˆ1 ⊗ eˆ 3 ⊗ eˆ1 ) +
=
+ A 211 ( eˆ 2 ⊗ eˆ1 ⊗ eˆ1 ) + A 221 ( eˆ 2 ⊗ eˆ 2 ⊗ eˆ1 ) + A 231 ( eˆ 2 ⊗ eˆ 3 ⊗ eˆ1 ) +
+ A 311 ( eˆ 3 ⊗ eˆ1 ⊗ eˆ1 ) + A 321 ( eˆ 3 ⊗ eˆ 2 ⊗ eˆ1 ) + A 331 ( eˆ 3 ⊗ eˆ 3 ⊗ eˆ1 ) +
+ A 112 ( eˆ1 ⊗ eˆ1 ⊗ eˆ 2 ) + A 122 ( eˆ1 ⊗ eˆ 2 ⊗ eˆ 2 ) + ...
Alternatively, this could have been written as:

A = A 111 eˆ1eˆ1eˆ1 + A 121 eˆ1eˆ 2eˆ1 + A 131 eˆ1eˆ 3eˆ1 +


+ A 211 eˆ 2eˆ1eˆ1 + A 221 eˆ 2eˆ 2eˆ1 + A 231 eˆ 2eˆ 3eˆ1 +
+ A 311 eˆ 3eˆ1eˆ1 + A 321 eˆ 3eˆ 2eˆ1 + A 331 eˆ 3eˆ 3eˆ1 +
+ A 112 eˆ1eˆ1eˆ 2 + A 122 eˆ1eˆ 2eˆ 2 + ...

14
Tensor Bases – 3rd ORDER TENSOR

A A 111 ( eˆ1 ⊗ eˆ1 ⊗ eˆ1 ) + A 121 ( eˆ1 ⊗ eˆ 2 ⊗ eˆ1 ) + A 131 ( eˆ1 ⊗ eˆ 3 ⊗ eˆ1 ) +
=
+ A 211 ( eˆ 2 ⊗ eˆ1 ⊗ eˆ1 ) + A 221 ( eˆ 2 ⊗ eˆ 2 ⊗ eˆ1 ) + A 231 ( eˆ 2 ⊗ eˆ 3 ⊗ eˆ1 ) +
A 311 ( eˆ 3 ⊗ eˆ1 ⊗ eˆ1 ) + A 321 ( eˆ 3 ⊗ eˆ 2 ⊗ eˆ1 ) + A 331 ( eˆ 3 ⊗ eˆ 3 ⊗ eˆ1 ) +
+
+ A 112 ( eˆ1 ⊗ eˆ1 ⊗ eˆ 2 ) + A 122 ( eˆ1 ⊗ eˆ 2 ⊗ eˆ 2 ) + ...
 In matrix notation:
 A  
  A 112113 AA 123 A 133  
A 132 
 A
 111 A 122 A 
 A A 213 AA 223 AA 233  
121 131

[A ] =  A211212 A 221222 A 231232   
  A A 313 AA 323 AA 333  
 A311312 A 321322 A 331332   
   

15
Tensor Bases – 3rd ORDER TENSOR

A A 111 ( eˆ1 ⊗ eˆ1 ⊗ eˆ1 ) + A 121 ( eˆ1 ⊗ eˆ 2 ⊗ eˆ1 ) + A 131 ( eˆ1 ⊗ eˆ 3 ⊗ eˆ1 )
=
+ A 211 ( eˆ 2 ⊗ eˆ1 ⊗ eˆ1 ) + A 221 ( eˆ 2 ⊗ eˆ 2 ⊗ eˆ1 ) + A 231 ( eˆ 2 ⊗ eˆ 3 ⊗ eˆ1 ) +
+ A 311 ( eˆ 3 ⊗ eˆ1 ⊗ eˆ1 ) + A 321 ( eˆ 3 ⊗ eˆ 2 ⊗ eˆ1 ) + A 331 ( eˆ 3 ⊗ eˆ 3 ⊗ eˆ1 ) +
+ A 112 ( eˆ1 ⊗ eˆ1 ⊗ eˆ 2 ) + A 122 ( eˆ1 ⊗ eˆ 2 ⊗ eˆ 2 ) + ...
 In index notation:
=A ∑ A ( eˆ ⊗ eˆ =
ijk
ijk ⊗ eˆ )
i j k

= A ijk ( eˆ i ⊗ eˆ j ⊗ eˆ k ) ≡ A ijk eˆ i eˆ j eˆ k
tensor as a
physical entity

[A ]ijk = A ijk component ijk of the tensor


in the given basis i, j , k ∈ {1, 2,3}

16
Repeated-index (or Einstein’s) Notation
 The Einstein Summation Convention: repeated Roman indices are
summed over. 3
i is a mute
index
ai bi = ∑ ai bi = a1b1 + a2b2 + a3b3
i =1

i is a talking 3
index and j is a Aij b j = ∑ Aij b j = Ai1b1 + Ai 2b2 + Ai 3b3
mute index j =1
 A “MUTE” (or DUMMY) INDEX is an index that does not appear in a
monomial after the summation is carried out (it can be arbitrarily changed
of “name”).
 A “TALKING” INDEX is an index that is not repeated in the same
monomial and is transmitted outside of it (it cannot be arbitrarily changed
of “name”). REMARK
An index can only appear up
to two times in a monomial.

18
Repeated-index (or Einstein’s) Notation

Rules of this notation:


1. Sum over all repeated indices.
2. Increment all unique indices fully at least once, covering all
combinations.

3. Increment repeated indices first.


4. A comma indicates differentiation, with respect to coordinate xi .
∂ui 3
∂ui ∂ 2ui 3
∂ 2ui ∂Aij 3 ∂A
u= = ∑ = ui , jj = ∑ 2 A= = ∑ ij

∂xi i =1 ∂xi ∂x j ∂x j j =1 ∂x j ∂x j j =1 ∂x j
i ,i ij , j

5. The number of talking indices indicates the order of the tensor result

19
Kronecker Delta δ
 The Kronecker delta δij is defined as:
1 if i = j
δ ij = 
0 if i ≠ j
 Both i and j may take on any value in
 Only for the three possible cases where i = j is δij non-zero.

1 if= i j (δ= δ= δ= 1)


δ ij =  11 22 33

0 if i ≠ j (δ12 =
δ13 =
δ 21... = 0)
REMARK

δ ij = δ ji Following Einsten’s notation: δ ii = δ11 + δ 22 + δ 33 = 3
Kronecker delta serves as a replacement operator:
=δ ij u j u=
i , δ ij A jk Aik
20
Levi-Civita Epsilon (permutation) e
 The Levi-Civita epsilon eijk is defined as:

 0 if there is a repeated index



eijk =
 +1 if ijk =
123, 231 or 312
 −1 if ijk =
213, 132 or 321

 3 indices 27 possible combinations.



eijk = −eikj
REMARK
The Levi-Civita symbol is also named
permutation or alternating symbol.

21
Example
 Prove the following expression is true:
eijk eijk = 6

23
Example - Solution
k =1 k =2 k =3
eijk eijk = e111e111 + e112e112 + e113e113 + j =1
=1
+ e121e121 + e122e122 + e123e123 + j = 2
i =1 = −1
+ e131e131 + e132e132 + e133e133 + j = 3
= −1
+ e211e211 + e212e212 + e213e213 +
i=2 + e221e221 + e222e222 + e223e223 +
=1
+ e231e231 + e232e232 + e233e233 +
=1
+ e311e311 + e312e312 + e313e313 +
= −1
i=3 + e321e321 + e322e322 + e323e323 +

+ e331e331 + e332e332 + e333e333 = 6

24
Vector Operations
Tensor Algebra

25
Vector Operations
 Sum and Subtraction. Parallelogram law.

a+b = b+a = c c=
i ai + bi
a−b =d d=
i ai − bi

 Scalar multiplication

αa =
b=α a1eˆ1 + α a2eˆ 2 + α a3eˆ 3 bi = α ai

26
Vector Operations
 Scalar or dot product yields a scalar
where θ is the angle
u⋅v =u v cos θ between the vectors u and v

 In index notation:
 i =3

u ⋅ v= ui eˆ i ⋅ v j eˆ j= ui v j eˆ i ⋅ eˆ j= ui v jδ ij= ui v i = ∑ i i = [ u] [ v ]
T
u v
 i =1 
u v δ ij =  0 (i ≠ j )

= 1=
 Norm of a vector ( j i)

u =(u ⋅ u ) =( uiui )
12 12

u = u ⋅ u = ui eˆ i ⋅ u j eˆ j = ui u jδ ij = ui ui
2

27
Vector Operations
 Some properties of the scalar or dot product

u ⋅ v = v ⋅u
u⋅0 = 0
u ⋅ (α v + β w )= α ( u ⋅ v ) + β ( u ⋅ w ) Linear operator

u ⋅u > 0 u≠0
=
u ⋅u 0 = u 0
u=⋅ v 0, u ≠ 0, v ≠ 0 u⊥v

28
Vector Operations
 Vector product (or cross product ) yields another vector
c =a × b =−b × a
c = a b sin θ where θ is the angle
between the vectors a and b
0 ≤θ ≤π
 In index notation:
=
c c= ˆ eijk a j bk eˆ i
i ei ⇒ =
ci eijk a j bk i ∈ 
i =1 i=2 i=3
 eˆ1 eˆ 2 eˆ 3 
= det  a1 a2 a3 
symb
c=( a2b3 − a3b2 ) eˆ1 + ( a3b1 − a1b3 ) eˆ 2 + ( a1b2 − a2b1 ) eˆ 3
a2b3 + e132 a3b2 e231 a3b1 + e213 a1b3 e312 a1b2 + e321 a2b1  b1 b2 b3 
e
123
     
=1 =−1 =1 = −1 =1 = −1

29
Vector Operations
 Some properties of the vector or cross product

u × v =− ( v × u )
u ×=
v 0, u ≠ 0, v ≠ 0 u || v
u × ( a v + b w ) = au × v + b u × w Linear operator

30
Vector Operations
 Tensor product (or open or dyadic product) of two vectors:
A = u ⊗ v ≡ uv
Also known as the dyad of the vectors u and v, which results in a 2nd
order tensor A.
 Deriving the tensor product along an orthonormal basis {êi}:
A = ( u ⊗ v ) = ( ui eˆ i ) ⊗ ( v j eˆ j ) = ui v j ( eˆ i ⊗ eˆ j ) = Aij ( eˆ i ⊗ eˆ j )

[ A ]ij =Aij =[ u ⊗ v ]ij =ui v j i, j ∈ 


 In matrix notation:
 u1   u1 v1 u1 v 2 u1 v 3   A11 A12 A13 
[u ⊗= v ] u2  [ v1
v ] [u ][= v 3 ] u2 v1 u2 v 2
= u2 v 3  =  A21 A23 
T
v2 A22
u3  u3 v1 u3 v 2 u3 v 3   A31 A32 A33 

31
Vector Operations
 Some properties of the open product:
(u ⊗ v ) ≠ ( v ⊗ u )
( u ⊗ v ) ⋅ w =u ⊗ ( v ⋅ w ) =u ( v ⋅ w ) =( v ⋅ w ) u
u ⊗ (α v + β w ) = α u ⊗ v + β u ⊗ w Linear operator
( u ⊗ v )( w ⊗ x ) = ( u ⊗ x )( v ⋅ w )
u ⋅ ( v ⊗ w ) = (u ⋅ v ) ⊗ w = (u ⋅ v ) w = w (u ⋅ v )

32
Example
 Prove the following property of the tensor product is true:
u ⋅ ( v ⊗ w ) = (u ⋅ v ) ⊗ w

33
Tensor Operations
Tensor Algebra

39
Tensor Operations
 Summation (only for equal order tensors)
A+B =B+A =C C=
ij Aij + Bij

 Scalar multiplication (scalar times tensor)


αA = C Cij = α Aij

40
Tensor Operations
 Dot product (.) or single index contraction product

A ⋅ b =
c ci = Aij b j Index “j” disappears (index
2nd 1st 1st contraction)
order order order

A ⋅ b =C Cij = Aijk bk Index “k” disappears (index


2nd
contraction)
3rd 1st
order order order

A ⋅ B =
C Cik = Aij B jk Index “j” disappears (index
2nd 2nd 2nd contraction)
order order order
REMARK
A⋅B ≠ B⋅A A⋅A =
A2

41
Tensor Operations
 Some properties:
A ⋅ (α b + β c )= α A ⋅ b + β A ⋅ c Linear operator

 2nd order unit (or identity) tensor

1⋅ u = u ⋅1 = u
1 = δ ij e j ⊗ ei = ei ⊗ ei 1 0 0 
 [1] = 0 1 0 
[1]ij = δ ij
 0 0 1 

42
2nd Order Tensor Operations
 Some properties:

1⋅ A = A = A ⋅1
A ⋅ ( B + C) = A ⋅ B + A ⋅ C
A ⋅ ( B ⋅ C ) =( A ⋅ B ) ⋅ C = A ⋅ B ⋅ C
A⋅B ≠ B⋅A

43
Example
 When does the relation n ⋅ T = T ⋅ n hold true ?

44
2nd Order Tensor Operations
 Transpose
 A11 A12 A13 
[ A ] = A21 A22 A23  [ AT ]
 A11
A
A21
A22
A31 
A32 
(A )
T T
=A
 12 ( A ⋅ B ) =BT ⋅ AT
T

 A31 A32 A33   A13 A23 A33 


( ⊗ ) =⊗
T
u v v u
[ AT ]ij = A ji (α A + β B )
T
=α AT + β BT
 Trace yields a scalar
Tr ( A ) = Aii ( = A11 + A22 + A33 ) Tr ( a ⊗ b ) =Tr  ai b j  =ai bi =a ⋅ b
 Some properties:
Tr ( AT ) = Tr A Tr ( A + B ) = TrA + TrB
Tr (α A ) = α TrA Tr ( A ⋅ B )= Tr ( B ⋅ A )
48
2nd Order Tensor Operations

 Double index contraction or double (vertical) dot product (:)


A : B = c c = Aij Bij Indices “i,j” disappear (double index
zero contraction)
2nd 2nd
order order order
(scalar)
A : B = c ci = Aijk B jk Indices “j,k” disappear (double index
contraction)
3rd 2nd 1st
order order order

A : B = C Cij =  ijkl Bkl Indices “k,l” disappear (double index


4th 2nd 2nd contraction)
order order order
 Indices contiguous to the double-dot (:) operator get vertically repeated
(contraction) and they disappear in the resulting tensor (4 order reduction of the sum
of orders).

49
2nd Order Tensor Operations
 Some properties

A : B= Tr ( AT ⋅ B )= Tr ( BT ⋅ A )= Tr ( A ⋅ BT )= Tr ( B ⋅ AT )= B : A

1= =
: A Tr A A :1
A : ( B ⋅ C ) = ( BT ⋅ A ) : C = ( A ⋅ CT ) : B
A : ( u ⊗ v ) =u ⋅ ( A ⋅ v )
(u ⊗ v ) : ( w ⊗ x) = (u ⋅ w ) ⋅ ( v ⋅ x)

REMARK
A : B= C : B ⇒ A= C

50
2nd Order Tensor Operations

 Double index contraction or double (horizontal) dot product (··)


A ⋅⋅ B = c c = Aij B ji Indices “i,j” disappear (double index
zero contraction)
2nd 2nd
order order order
(scalar)

A ⋅⋅ B = c ci = Aijk Bkj Indices “j,k” disappear (double index


contraction)
3rd 2nd 1st
order order order

A ⋅⋅ B = C Cij =  ijkl Blk Indices “k,l” disappear (double index


4th 2nd 2nd contraction)
order order order
 Indices contiguous to the double-dot (··) operator get horizontally repeated
(contraction) and they disappear in the resulting tensor (4 orders reduction of the sum
of orders).

51
Tensor Operations
 Norm of a tensor is a non-negative real number defined by
(A : A) ( Aij Aij )
12
= = ≥0
12
A

A ⋅⋅ B= Tr ( A ⋅ B )= Tr ( B ⋅ A )= B ⋅⋅A

REMARK
1 ⋅⋅A= TrA= A ⋅⋅1
A : B ≠ A ⋅⋅B
Unless one of the two
tensors is symmetric.

52
Example
 Prove that:

A : B Tr ( AT ⋅ B )
=

⋅⋅ B Tr ( A ⋅ B )
A=

53
2nd Order Tensor Operations
 Determinant yields a scalar
 A11 A12 A13 
[ A ] det  A21 A23  eijk A1i A
1
= =
det A det A22 = = A
2 j 3k eijk epqr Api Aqj Ark
6
A A32 A33 
 Some properties:  31
REMARK
det ( A ⋅ B=
) det A ⋅ det B The tensor A is SINGULAR if and
det AT = det A only if det A = 0.
det (α A ) = α 3 det A A is NONSINGULAR if det A ≠ 0.
 Inverse
There exists a unique inverse A-1 of A when A is nonsingular, which
satisfies the reciprocal relation:

 A ⋅ A −1 ==
1 A −1 ⋅ A

 A=
−1
ik Akj A= −1
ik Akj δ ij i, j , k ∈ {1, 2,3}
55
Example
 Prove that det A = eijk A1i A2 j A3.k

57
Differential Operators
Tensor Algebra

64
Differential Operators
 A differential operator is a mapping that transforms a field
v ( x ) , A ( x ) ... into another field by means of partial derivatives.
 The mapping is typically understood to be linear.
 Examples:
 Nabla operator
 Gradient
 Divergence
 Rotation
 …

65
Nabla Operator
 The Nabla operator  is a differential operator
“symbolically” defined as:
symbolic
∂ symb. ∂
∇= = eˆ i
∂x ∂xi
 In Cartesian coordinates, it can be used as a (symbolic) vector
on its own:
 ∂ 
 

 1x
symb. 
∂ 
[∇ ] = 
 ∂x2 
 ∂ 
 

 3
x

66
Gradient
 The gradient (or open product of Nabla) is a differential
operator defined as:
 Gradient of a scalar field Φ(x):
 Yields a vector
 symb.
∂ ∂Φ
[ ∇Φ ]i [
= ∇ ⊗ Φ ]i [ ]i
= ∇ Φ =
∂xi
Φ =
∂xi
i ∈ {1, 2,3}

 ∂Φ
∇Φ = [∇Φ ] eˆ = ∂Φ eˆ ∇Φ = eˆ i
 i i
∂xi
i
∂xi
 Gradient of a vector field v(x):
 Yields a 2nd order tensor
 symb.
∂ ∂v j
 [ ∇ ⊗ v ] = [ ∇ ] [ v ] = vj = i, j ∈ {1, 2,3}

ij i j
∂xi ∂xi
 ∂v j
∇v = ∇ ⊗ v = [∇ ⊗ v ] eˆ ⊗ eˆ = ∂v j =
∇v eˆ i ⊗ eˆ j
eˆ i ⊗ eˆ j ∂xi
 ij i j
∂xi
67
Gradient
 Gradient of a 2nd order tensor field A(x):
 Yields a 3rd order tensor

 symb.
∂ ∂A jk
[∇A ]ijk = [∇ ⊗ A ]ijk = [∇ ]i [ A ] jk = A jk = i, j , k ∈ {1, 2,3}
 ∂xi ∂xi

∇A = ∇ ⊗ A = [∇ ⊗ A ] eˆ ⊗ eˆ ⊗ eˆ = ∂A jk eˆ ⊗ eˆ ⊗ eˆ
 ijk i j k
∂xi
i j k

∂A jk
=
∇A eˆ i ⊗ eˆ j ⊗ eˆ k
∂xi

68
Divergence
 The divergence (or dot product of Nabla) is a differential
operator defined as :
 Divergence of a vector field v(x):
 Yields a scalar
symb.
∂ ∂vi ∂vi
∇ ⋅ v = [∇ ]i [ v ]i = vi = ∇⋅v =
∂xi ∂xi ∂xi
 Divergence of a 2nd order tensor A(x):
 Yields a vector
 symb.
∂ ∂A ij
 ∇ ⋅ A j = [∇ ]i [ A ]ij =
[ ] A ij = j ∈ {1, 2,3}
 ∂x i ∂xi

∇ ⋅ A = [∇ ⋅ A ] eˆ = ∂A ij eˆ ∂A ij
 j j
∂xi
j
∇⋅A = eˆ j
∂xi

69
Divergence

 The divergence can only be performed on tensors of order 1 or


higher.

 If ∇ ⋅ v = 0, the vector field v ( x ) is said to be solenoid (or


divergence-free).

70
Rotation
 The rotation or curl (or vector product of Nabla) is a differential
operator defined as:
 Rotation of a vector field v(x):
 Yields a vector
[ symb. symb.
∂ ∂ vk
 ∇ × v=]i e [ ∇ ] [ v ]
= e =
v e i ∈ {1, 2,3}
∂ ∂x j
ijk j k ijk k ijk
 x j
 ∂ vk
∇ × v = [∇ × v ]i eˆ = e ∂ v k eˆ ∇ × v = eijk eˆ i
 i ijk
∂x j
i
∂x j
 Rotation of a 2nd order tensor A(x):
 Yields a 2nd order tensor
[ symb.
∂ ∂ A kl
=
]
∇ × A il e= A kl eijk i, j , k ∈ {1, 2,3}
∂ ∂
ijk
 x j x j
 ∂ A kl
∇ × A = [∇ × A ]il eˆ ⊗ eˆ = e ∂ A kl eˆ ⊗ eˆ = ∇ × A eijk
∂x j
eˆ i ⊗ eˆ l
 i l ijk
∂x j
i l

71
Rotation

 The rotation can only be performed on tensors of order 1 or


higher.

 If ∇ × v = 0 , the vector field v ( x ) is said to be irrotational (or


curl-free).

72
Differential Operators - Summary

scalar field vector field 2nd order tensor


Φ(x) v(x) A(x)

[∇ ⊗ Φ ]i = [∇ ⊗ v ]ij = [∇ ⊗ A ]ijk =
GRADIENT ∂Φ ∂v ∂A jk
= [∇Φ ]i = [∇v ]ij = j
= [ ]ijk
=∇ A =
∂xi ∂xi ∂xi

∂vi ∂A ij
DIVERGENCE ∇⋅v = [∇ ⋅ A ] j =
∂xi ∂xi

ROTATION [∇ × v ]i = eijk ∂ v k [∇ × A ]il = eijk ∂ A kl


∂x j ∂x j

73
Example
 Given the vector v =v ( x ) =x1 x2 x3eˆ1 + x1 x2eˆ 2 + x1eˆ 3
determine ∇ ⋅ v, ∇ × v, ∇v.

74
Example - Solution
 x1 x2 x3 
v =v ( x ) =x1 x2 x3eˆ1 + x1 x2eˆ 2 + x1eˆ 3
[ v ] =  x1 x2 
 x1 
 Divergence:

∂vi
∇⋅v =
∂xi

∂vi ∂v1 ∂v 2 ∂v3


∇=
⋅v = + + = x2 x3 + x1
∂xi ∂x1 ∂x2 ∂x3

75
Example - Solution
 Divergence:  x1 x2 x3 
∇⋅v =
∂vi [ v ] =  x1 x2 
∂xi
 x1 
 In matrix notation:
 x1 x2 x3  symb
T
 ∂ ∂ ∂  
,   x1 x2  =
symb symb
∇ ⋅ v = [∇ ] [ v ] =  ,
T

 ∂x1 ∂x2 ∂x3   x 


 1 
13
31
symb
∂ ∂ ∂ ∂ ( x1 x2 x3 ) ∂ ( x1 x2 ) ∂x1
= x1 x2 x3 + x1 x2 + x1 = + + = x2 x3 + x1
∂x1 ∂x2 ∂x3 ∂x1 ∂x2 ∂x3
11

76
Integral Theorems
Tensor Algebra

81
Divergence or Gauss Theorem
 Given a field A in a volume V with closed boundary surface
∂V and unit outward normal to the boundary n , the
Divergence (or Gauss) Theorem states:

∫ ∇ ⋅A
V
=
dV ∫ ∂V
n ⋅A dS

∫ A ⋅∇=
V
dV ∫
∂V
A ⋅ n dS
Where:
A represents either a vector field ( v(x) ) or a tensor field ( A(x) ).

82
Generalized Divergence Theorem
 Given a field A in a volume V with closed boundary surface
∂V and unit outward normal to the boundary n , the
Generalized Divergence Theorem states:

∫ ∇ ∗A
V
=
dV ∫∂V
n ∗A dS

∫ A ∗∇=
V
dV ∫
∂V
A ∗ n dS
Where:
 ∗ represents either the dot product ( · ), the cross product (  ) or the
tensor product (  ).
 A represents either a scalar field ( ϕ(x) ), a vector field ( v(x) ) or a
tensor field ( A(x) ).

83
Example
 Use the Generalized Divergence Theorem to show that

∫S
xi n j dS = V δ ij
nj
where xi is the position vector of nj .

∫∂V
A ∗ n=
dS ∫ A ∗∇ dV
V

85
References
Tensor Algebra

88
References

 José Mª Goicolea, Mecánica de Medios Continuos: Resumen de Álgebra y


Cálculo Tensorial, UPM.

 Eduardo W. V. Chaves, Mecánica del Medio Continuo,Vol. 1 Conceptos


básicos, Capítulo 1: Tensores de Mecánica del Medio Continuo, CIMNE,
2007.

 L. E. Malvern. Introduction to the mechanics of a continuous medium.


Prentice-Hall, Englewood Clis, NJ, 1969.

 G. A. Holzapfel. Nonlinear solid mechanics: a continuum approach for


engineering. 2000.

89

Vous aimerez peut-être aussi