Mirzaei ElasticityLecture PDF

M.
Mirzaei, Elasticity
-1-
Theory of
Elasticity
by
Majid Mirzaei
These lecture notes have been prepared as supplementary material for a
graduate level course on Theory of Elasticity. You are welcome to read
or print them for your own personal use. All other rights are reserved.
No originality is claimed for these notes other than selection,
organization, and presentation of the material.
The major references that have been used for preparation of the lecture
notes are as follows:
Timoshenko & Goodier, "Theory of Elasticity"

Boresi & Chong, "Elasticity in Engineering Mechanics"
Phillip L. Gould, "Introduction to Linear Elasticity"
Provan, J.W., "Stress Analysis Lecture Notes"
M. Mirzaei, Elasticity
-2-
Theory of Elasticity
Lecture Notes
Majid Mirzaei, PhD
Associate Professor
Dept. of Mechanical Eng., TMU
mmirzaei@modares.ac.ir
http://www.modares.ac.ir/eng/mmirzaei
Chapter 1
Field Equations of Linear Elasticity
In many engineering applications a pointwise description of the field quantities, such as
stress and strain, is required throughout the body. The distinctive feature of the theory of
elasticity, compared to the alternative approaches like the strength of materials, is that it
provides a consistent set of equations which may be solved (using the advanced
techniques of applied mathematics) to obtain a unique pointwise description of the
distribution of the stresses, strains, and displacements for a particular loading and
geometry.
Tensors, Indicial Notation

In the theory of elasticity we usually deal with physical quantities which are independent
of any particular coordinate system that may be used to describe them. At the same time,
these quantities (tensors) are usually specified in a particular coordinate system by their
components. Specifying the components of a tensor in one coordinate system determines
the components in any other system. The physical laws of Elasticity are in turn expressed
by tensor equations. Because tensor transformations are linear and homogeneous, such
tensor equations, if they are valid in one coordinate system, are valid in any other
coordinate system. This invariance of tensor equations under a coordinate transformation
is one of the principal reasons for the usefulness of tensor methods in the theory of
Elasticity.
In a three-dimensional Euclidean space (such as ordinary physical space) the number of
components of a tensor is 3N, where N is the order of the tensor. Accordingly a tensor of
order zero is specified in any coordinate system in three-dimensional space by one
component. Tensors of order zero are called scalars. Physical quantities having
magnitude only are represented by scalars. Tensors of order one have three coordinate
components in physical space and are known as vectors. Quantities possessing both
magnitude and direction are represented by vectors. Second-order tensors are called
dyadics. Higher order tensors such as triadics, or tensors of order three, and tetradics, or
tensors of order four can also be defined.
-3-
In these notes we use the indicial notation for presentation of tensorial quantities and
equations. Under the rules of indicial notation, a letter index may occur either once or
twice in a given term. When an index occurs unrepeated in a term, that index is
understood to take on the values 1,2,.. .,N where N is a specified integer that determines
the range of the index. Unrepeated indices are known as free indices. The tensorial rank
of a given term is equal to the number of free indices appearing in that term. Also,
correctly written tensor equations have the same letters as free indices in every term. In
ordinary physical space a basis is composed of three, non-coplanar vectors, and so any
vector in this space is completely specified by its three components. Therefore the range
on the index of ai (which represents a vector in physical space) is 1,2,3. Accordingly the
symbol ai is understood to represent the three components a1, a2, a3. For a range of three
on both indices, the symbol Aij represents nine components (of the second-order tensor
(dyadic) A).
When an index appears twice in a term, that index is understood to take on all the values
of its range, and the resulting terms summed. In this so-called summation convention,
repeated indices are often referred to as dummy indices because their replacement by any
other letter not appearing as a free index does not change the meaning of the term in
which they occur. In general, no index occurs more than twice in a properly written
term.
Kinematics of Deformable Solids

Fig. 1.1 shows the undeformed configuration of a material continuum at time t = t0
together with the deformed configuration at a later time t = t. It is useful to refer the
initial and final configurations to separate coordinate systems as shown in the figure.
Deformed
Configuration
t=t
C
undeformed
Configuration
t=t0
C0
X2
X3
u
x2
R0
x1
b
x3
X1
Fig. 1.1
-4-
Suppose a material point is at position X in the undeformed solid, and moves to a

position x when the solid is loaded. We may describe the deformation and motion of a
solid by a mapping in the following form:
x = ( X, t )
(1-1)
Now we consider the two coordinate systems to be coincident, as shown in Fig. 1.2. The
displacement of a material point is:
u(t ) = x(t ) X
(1-2)
C0
X2 x2
X3 x3
R0 X
X1
x1
b C
Deformed
Configuration
undeformed
Configuration
Fig. 1.2
Next, consider two straight parallel lines on the reference configuration of a solid (see
Fig. 1.3). If the deformation of the solid is homogeneous, the two lines remain straight in
the deformed configuration, and the lines remain parallel.
ds1
dS1
dS2
ds 2
undeformed
Configuration
Fig. 1.3
Deformed
Configuration
-5-
Furthermore, the lines stretch by the same amount, i.e.:

(1-3)
ds1 ds2
=
dS1 dS2
The difference (ds ) 2 (dS ) 2 is used to define a measure of deformation, which occurs in
the vicinity of the particles between the initial and final configurations.
(1-4)
(ds ) 2 = dx12 + dx22 + dx32 = dxi dxi = dx dx

(dS ) 2 = dX 12 + dX 22 + dX 32 = dX i dX i = dX dX
u+du
dS
undeformed
Configuration
ds
Deformed
Configuration
Fig. 1.4
Referring to Fig. 1.4, we may further write:
dS + u + du = u + ds, du = ds dS
(1-5)
We also have,
(ds) 2 (dS ) 2 = dxi dxi dX i dX i
(1-6)
and,
xi = xi ( X 1 , X 2 , X 3 )
dxi =
xi
x
x
dX 1 + i dX 2 + i dX 3
X 1
X 2
X 3
dxi = xi , j dX j = F dX
(1-7)
-6-
where the comma denotes differentiation with respect to a spatial coordinate. In the
above F is known as deformation gradient tensor. Next, we define
u = u1e x1 + u2e x2 + u3e x3 = ui e xi
(1-8)
and proceed as follows:

(ds ) 2 (dS ) 2 = xi , j dX j xi , k dX k dX i dX i = (F dX) (F dX) dX dX
= ( xi , j xi , k ij ik ) dX j dX k
= ( X i + ui ), j ( X i + ui ), k jk dX j dX k
= ( ij + ui , j ) ( ik + ui , k ) jk dX j dX k
(1-9)
= jk + u j , k + uk , j + ui , j ui ,k jk dX j dX k
= ( u j ,k + uk , j + ui , j ui ,k ) dX j dX k
Rewriting the indices in a more proper form, we get

(ds ) 2 (dS ) 2 = ( ui , j + u j ,i + uk ,i uk , j ) dX i dX j
(1-10)
= 2 ijL dX i dX j
in which we distinguish the Lagrangian strain tensor ijL for characterization of the
deformation near a point.
ijL =
1
( ui, j + u j ,i + uk ,iuk , j )
2
(1-11)
Alternatively we may write

X i = X i ( x1 , x2 , x3 )
dX i =
X i
X
X
dx1 + i dx2 + i dx3
x1
x2
x3
dX i = X i , j dx j = F dx
and proceed as follows:
(1-12)
-7-
(ds ) 2 (dS ) 2 = dxi dxi X i , j dx j X i ,k dxk = dx dx (F dx) (F dx)

= ( ij ik X i , j X i ,k ) dx j dxk
(1-13)
= jk ( xi ui ), j ( xi ui ),k dx j dxk
= jk ( ij ui , j ) ( ik ui , k ) dx j dxk
= jk jk + uk , j + u j , k ui , j ui ,k dx j dxk
= ( u j ,k + uk , j ui , j ui ,k ) dx j dxk
Replacing the dummy indices i and k, we have

(ds ) 2 (dS ) 2 = ( ui , j + u j ,i uk ,i uk , j ) dxi dx j
= 2 ijE dxi dx j
(1-14)
This time, we define the Eulerian strain tensor ijL to characterize the deformation near a
point.
ijE =
(1-15)
1
( ui, j + u j ,i uk ,iuk , j )
2
In practice, we need to make a number of assumptions to simplify the equations of linear

elasticity. A major one is to assume that deformations are infinitesimal. In most practical
circumstances it is sufficient to assume uk ,i uk , j << 1
We use this to define a linear measure of deformation in linear elasticity, and define the
infinitesimal strain tensor as:
ijL = ijE = ij =
1
ui , j + u j ,i )
(
2
(1 16)
In order to check the relationship between the above definition of strain and the
conventional definition, we consider a one-dimensional case as follows:
(ds ) 2 (dS ) 2 = 2 ijL dX i dX j = 2 22 dX i dX i = 2 22 (dS ) 2
(ds ) 2 (dS ) 2 (ds + dS )(ds dS ) ds dS
22 =
=
=
2(dS ) 2
dS (ds + dS )
dS
(1-17)
-8-
The linear strain tensor in Cylindrical and Spherical

Coordinate Systems
Cylindrical
ur
rr = r
1 ur u u
+
r
r
r
1 u ur
=
+
r
r
2 r =
ur u z
+
z
r
u 1 u z
2 z = +
z r
u
zz = z
z
2 rz =
(1-18)
Spherical
ur
rr =
r
2 r =
1 ur u u
+
r
r
r
1 u ur
=
+
r
r
1 ur u u
+
r sin r
r
1 u 1 u u cot
+
2 =
r sin r
r
1 u ur u cot
+ +
=
r sin r
r
2 r =
(1-19)
Compatibility
Conditions of compatibility, imposed on the components of strain, are necessary and
sufficient to insure a continuous single-valued displacement field. The procedure is to
eliminate the displacements between kinematic equations to produce equations with only
strain as unknowns. For instance we may write:
ll ,mm = ul ,lmm
mm ,ll = um ,mll
1
lm ,lm = (ul ,mlm + um ,llm )
2
1
= (ul ,lmm + um ,mll )
2
2 lm,lm = ll ,mm + mm ,ll
(1-20)
-9-
In general, we have
ij ,kl + kl ,ij = ik , jl + jl ,ik
(1-21)
which represents 81 equations, only six of them are independent.
11,22 + 22,11 = 212,12

22,33 + 33,22 = 2 23,23
33,11 + 11,33 = 2 31,31
12,13 + 13,12 23,11 = 11,23
(1-22)
23,21 + 21,23 31,22 = 22,31

31,32 + 32,31 12,33 = 33,12
Assignment 1:
Noting that = ( ),i ei and using indicial notation, prove the identities or find the right
hand side:
2 = = ,ii
( v ) = v + v
2 ( ) = 2 + 2 + 2
3 ( ) = 2 ( ) = ?
( ) = ?
- 10 -
Kinetics of Deformable Solids

In general, two types of forces can be applied to a solid body.
(i) External forces applied to its boundary (e.g. forces arising from contact with another
solid or fluid pressure).
(ii) Externally applied forces that are distributed throughout a body (e.g. gravitational,
magnetic, and inertial forces).
C0
X2
X1
X3
R0
b C
Deformed
Configuration
undeformed
Configuration
Fig. 1.5
The stress vector or surface traction t at a point represents the force acting on the
surface per unit area and can be defined as:
(1-23)
dP
dA0 dA
t = lim
in which dA is an element of area on a surface subjected to a force dP (see Fig. 1.6).
dP
dA
Fig. 1.6
The body force vector denotes the external force acting on the interior of a solid per unit
volume and can be defined as:
dP
dV 0 dV
b = lim
(1-24)
- 11 -
in which dV denotes an infinitesimal volume element subjected to a force dP (Fig. 1.7).
dP
dV
Fig. 1.7
Internal forces induced by external loading

The solid body shown in Fig. 1.8 is subjected to a balanced system of externally applied
forces. As a result, internal forces are developed in order to keep the body together.
Suppose that the body is cut into two parts. The force Pn represents the resultant force
that acts on the two faces in order to keep the two parts in load equilibrium.
n
t
Pn
b C
e2
-n
e1
e3
Fig. 1.8
The internal traction vector Tn represents the force per unit area acting on a plane with
normal vector n inside the deformed solid an can be defined as:
Tn = lim
dA0
dPn
dA
(1-25)
- 12 -
in which dA is an element of area in the interior of the solid, with normal n (see Fig. 1.9).
dPn
n
dA
Fig. 1.9
The components of Cauchy stress in a given basis can be visualized as the tractions
acting on planes with normals parallel to each basis vector, as depicted in Fig. 1.10.
e2
T2
22
21
23
12
32
T3
11
33
31
13
e1
e3
Fig. 1.10
T1
Here, we may write:

T1 = 11e1 + 12e 2 + 13e3
T2 = 21e1 + 22e 2 + 23e3
T3 = 31e1 + 32e 2 + 33e3
or Ti = ij e j
(1-26)
- 13 -
In order to find the components of the traction vector on an arbitrary plane (represented
by n) we impose the equilibrium on a tetrahedral element, as shown in Fig. 1.11.
e2
dAn
dA1
Tn dAn
e1
-e1
T(-e1) dA1
e3
Fig. 1.11
1
Tn dAn Ti dAi + f hdAn = 0,

3
h 0 Tn dAn Ti dAi = 0
dAi = dAn Cos(n, ei )
(1-27)
Tn dAn Ti dAnn ei = 0
Tn = Ti ei , Ti = ij e j , n ei = ni
Ti ei ij e j ni = 0 , Ti ei = ij e j ni = ji ei n j
Accordingly, we obtain the following expressions which are called: the stress boundary
equations
Ti = ji n j
(1-28)
Principle Stresses
For practical purposes, it is convenient to break down the traction vector into normal and
shear components as follows:
nn = Tn n = Ti ei n = Ti ni
= ji n j ni
(1-29)
- 14 -
and,
1
2 2
ns = (TT
i i nn )
(1-30)
In order to find the extremum values of the normal components of the stress tensor (the
principle stresses) we may write:
nn = ji n j ni
(1-31)
= ji n j ni (ni ni 1)
where is the lagrangian multiplier. Next, we write:
nn
= ji n j ni = 0
ni
= ji n j ij n j = 0
(1-32)
= ( ij ij )n j = 0
which implies that
11
12
13
21
22
23 = 0
31
31
33
(1-33)
The characteristic polynomial is
3 I I 2 + I II I III = 0
(1-34)
The roots of the above characteristic polynomial are the eigenvalues of our problem, or
the principle stresses. The invariants of the stress tensor are defined as:
I I = ii
I II = ( ii jj ij ji )
2
I III = ij
(1-35)
Finally, for each eigenvalue, we may find the eigenvectors which are the principle stress
directions.
- 15 -
1 ( ij ij )n j = 0 n11 , n12 , n31

2 ( ij ij )n j = 0 n12 , n22 , n32
(1-36)
3 ( ij ij )n j = 0 n , n , n
3
1
3
2
3
3
Assignment 2:
For the given stress tensor, determine the maximum shear stress and show that it acts in
the plane which bisects the maximum and minimum stress planes.
0
5 0
ij = 6 12

1
Conservation of Linear Momentum

In order to find the governing equations for distribution of the stress tensor within a body,
we start by considering an arbitrary material region with volume V and surface S. Based
on the general concept of conservation of linear momentum, we may write:
TdA + fdV = u&&dV

S
(1-37)
which, using (1-28), can be written as
ji
n j dA + f i dV = u&&i dV
V
(1-38)
Now we use the Divergence theorem of Gauss and write
G ndA = GdV
S
ji , j
(1-39)
dV + f i dV ( ji , j + f i ) dV = u&&i dV
V
Since we are dealing with an arbitrary volume, the three equations of motion can be
derived from (1-39) as
ji , j + fi = u&&i
(1 40)
- 16 -
The static form of the above equations (usually known as the equilibrium equations) is
ji , j + f i = 0
(1-41)
The Equations of Motion in Cylindrical and Spherical Coordinate Systems can be

written as:
rr 1 r rz rr
+
+
+
+ f r = u&&r
r
r
r
z
r 1 z 2 r
+
+
+
+ f = u&&
r
r
r
z
rz 1 z zz rz
+
+
+
+ f z = u&&z
r
r
r
z
(1-42)
and
rr 1 r
1 r 1
+
+
+ (2 rr r cot ) + f r = u&&r
r
r sin
r
r
r 1
1 1
+
+
+ [3 r + ( ) cot ] + f = u&&
r
r sin
r
r
r 1
1 1
+
+
+ (3 r + 2 cot ) + f = u&&
r
r sin
r
r
(1-43)
Conservation of Angular Momentum

Once again, let us consider a material region with volume V and surface S. Based on the
general concept of conservation of angular momentum we may write:
(X T)dA + (X f )dV = (X u&&)dV

S
ijk
X iT j dA + ijk X i f j dV = 0
V
which, using (1-28) and the Divergence theorem of Gauss, can be written as:
(1-44)
ijk
- 17 -
X i ( lj nl )dA + ijk X i f j dV = 0
V
ijk
( X i lj ),l dV + ijk X i f j dV = 0
V
ijk [( X i lj ),l + X i f j ]dV = 0
(1-45)
ijk
[ X i ,l lj + X i lj ,l + X i f j ]dV = 0
Next, we use (1-41) to simplify the above as:
ijk X i,l lj + X i lj ,l + f j dV = 0
=0
ijk il lj dV = 0
(1-46)
Since we are dealing with an arbitrary volume, we can write:
ijk ij = 0
(1-47)
ij = ji
which shows that the stress tensor is symmetric.
Constitutive Equations of Linear Elasticity

For an elastic body which is gradually strained at constant temperature, the components
of stress can be derived from the strain energy density , which is a quadratic function of
the strain components.
ij =
eij
(1-48)
Accordingly, we may write the most general form of the Hookes Law as:
ij = Eijkl ekl
(1-49)
- 18 -
in which Eijkl represents 81 components. Due to the symmetry of the stress and strain
tensors and also the elastic coefficient matrix, the number of independent elastic
constants reduces to 21.
11 E1111

22
33
=
12
23

31
E1122
E1133
2 E1112
2 E1123
E2222
E2233
2 E2212
2 E2223
E3333
2 E3312
2 E1212
2 E3323
2 E1223
2 E2323
2 E1131 11
2 E2231 22
2 E3331 33

2 E1231 12
2 E2331 23

2 E3131 31
(1-50)
which can be simplified to:

11 C11 C12

C22
22
33
=
12
23

31
C13 C14
C23 C24
C15
C25
C33
C34
C35
C44
C45
C55
C16 11
C26 22
C36 33

C46 12
C56 23

C66 31
(1-51)
The above expression represents the constitutive equations for an anisotropic elastic
material. Most engineering materials show some degree of isotropic behavior as they
possess properties of symmetry with respect to different planes or axes. We start with the
plane x2x3 as the plane of symmetry, which implies that the x1 axis can be reversed as
shown in Fig. 1.12.
x3
x3
x1
x2
x2
x1
Fig. 1.12: single plane symmetry
- 19 -
This corresponds to a coordinate transformation with the direction cosines shown in table
1.1.
axes
x1
x2
x1
-1
x2
x3
x3
Table 1.1
Using the transformation ij = ik jl kl , in which ik and jl are the direction cosines,
we find that:
11 = 11
12 = 12
= 22
22
= 23
23
33 = 33
31 = 31
(1-52)
Similarly, we can show that:
11 = 11
12 = 12
= 22
22
= 23
23
33 = 33
31 = 31
(1-53)
Hence, we may write:

11 11

22 22
33
33
=
=
12
12

23
23

31
31
which results in:
11

22

33

12
23

31
11

22
33
12
23
31
(1-54)
11 C11 C12

C22
22
33
=
12
23

31
- 20 -
C13
C14
C15
C23
C24
C25
C33
C34
C44
C35
C45
C55
C16 11
C26 22
C36 33

C46 12
C56 23

C66 31
(1-55)
As the constants should not change with the transformation, we must have C14, C16, C24,
C26, C34, C36, C45, and C56 = 0. Such materials are called monoclinic and need 13
constants to describe their elastic properties.
The double symmetry about the x1 and x2 axes will lead to further reduction of the
constants down to 9, which represents the orthotropic material (Note that for 2D
problems we need 4 elastic constants to describe the behavior of an orthotropic material).
axes
x1
x2
x1
-1
x2
x3
-1
x3
We may have an additional simplification by considering the directional independence in

elastic behavior by interchanging x2 with x3, and x1 with x2. Such materials are called
cubic with three independent constants.
axes
x1
x2
x3
axes
x1
1
x2
0
x3
0
x1
x2
x3
x1
0
x2
1
x3
0
Finally, we may consider rotational independence in material property and define the
constitutive equations for an isotropic elastic material with only two independent
constants:
axes
x1
x2
x3
x1
1
x2
0
x3
0
cos
sin
- sin
cos
11 2 +

22
33
=
12
23

31
- 21 -
2 +
2 +
0 11
0 22
0 33

0 12
0 23

2 31
(1-56)
in which and are called the Lam constants. The indicial form can be written as:
ij = 2 ij + ij kk
(1-57)
We may invert the above equation to express the strains in terms of stresses:
ij =
ij
ij kk
2
2 (2 + 3 )
(1-58)
The Lam constants are quite suitable from mathematical point of view, but they should
be related to the Engineering elastic constants obtained in the laboratory (like E and ) as
well. Table 1.2 shows the relationships between different elastic constants.
2
3
( 3 + 2 )
( + )
2 ( + )
E ,
E
(1 + )(1 2 )
,
2
(1 2 )
E,
( E 2 )
3 E
E
2 (1 + )
3k (1 2 )
2 (1 + )
E
3 (1 2 )
2 (1 + )
3 (1 2 )
E
3 ( 3 E )
2 (1 + )
3k (1 2 )
E
1
2
Table 1.2
k ,
3k
(1 + )
- 22 -
Formulation and Solution Methods of Elasticity Problems

In order to find the fifteen unknowns (three components of displacement, six components
of strain, and six components of stress) for an elasticity problem, we need fifteen
equations as follows:
ij =
(I)
1
( ui, j + u j ,i )
2
ji , j + f i = 0
ij = 2 ij + ij kk
ij =
(II)
or
ij
ij kk
2
2 (2 + 3 )
(III)
along with the compatibility and stress boundary equations:
ij ,kl + kl ,ij = ik , jl + jl ,ik
(IV)
Ti = ji n j
(V)
Classical Displacement Formulation

We start by substituting Eqs.(I) into the first form of Eqs.(III) to eliminate the strains:
ij = ij uk ,k + (ui , j + u j ,i )
(1-59)
Next, we substitute the above expression for the stresses in Eqs.(II) which resulys in:
ij uk ,ki + (ui , ji + u j ,ii ) + f j = 0

uk ,kj + (ui , ji + u j ,ii ) + f j = 0
(1-60)
Finally, using i for the dummy index in all terms we write:
u j ,ii + ( + )ui ,ij + f j = 0
(1-61)
- 23 -
or alternatively:
( + G )u j ,ij + Gui , jj + fi = 0
(1-62)
Equations (1-62) are three equilibrium equations in terms of displacements. They are
called Navier Equations and constitute the classical displacement formulation.
Navier Displacement Equations in Cylindrical and Spherical

Coordinates
Cylindrical:
I e
1 z
2G
+ fr = 0
r
z
r
I

( + 2G ) e 2G r z + f = 0
r
r
z
( + 2G )
( + 2G )
I e
(r ) r
2G
+ fz = 0
z

r
where
1 (rur ) 1 u u z
+
+
z
r r
r
u u
1 u z u
2r =
; 2 = r z
r
z
z
r
1 (ru ) ur
2 z =
r r

(1-63)
I e = eii = u =
- 24 -
Spherical:
I e
2G ( sin )
+ fr = 0
r r sin

I
2G 1 (r ) (r )
( + 2G ) e
+ f = 0
r
r
r sin
( + 2G )
( + 2G ) I e 2G (r ) (rr )
+ f = 0

r sin
r r
where
I e = eii = u =
(1-64)
1 ( r 2 ur )
1 (u sin )
1 u
+
+
r 2 r
r sin
r sin
1 (u sin ) u
1 ur 1 (ru )
; 2 =
r sin

r sin r r
1 (ru ) ur
2 =
r r

2r =
Classical Force Formulation

We start by substituting the RHS of the second form of Eqs. (III) for strains in Eqs. (IV)
to produce 81 stress compatibility equations as follows:
ij ,kl + kl ,ij ik , jl jl ,ik =
(
1 +
ij
tt , kl
kl tt ,ij ik tt , jl jl tt ,ik )
(1-65)
However, only six of the above equations are independent:
ij ,kk + kk ,ij ik , jk jk ,ik =
(
1 +
ij
tt , kk
kk tt ,ij ik tt , jk jk tt ,ik )
(1-66)
In general, the above expression represents nine equations with free indices i and j. Now
we substitute the equilibrium equations (II) in the above and simplify to have:
2 ij +
mm ,ij = fi , j + f j ,i +
ij f m,m
1 +
1
(1-67)
- 25 -
which are known as Beltrami-Michell compatibility relations and constitute the classical
force formulation.
Assignment 3:
For a hollow sphere under internal and external pressures find the stress, strain and
displacement distributions.
Beltrami-Michell Equations in Cylindrical Coordinates:

( fr )
1 rr 1 2 rr 2 rr
1 2 It
1 ( rf r ) 1 ( f ) ( f z )
+
+
=
+
+
2
r
+ 2
2
2
2
r r r r
r
1 + r
1 r r
z
z
r
2
2
2
( f )
1 1
1 It
1 ( rf r ) 1 ( f ) ( f z )
r
+ 2
+
+
=
+
+
2
2
2
r r r r
r
1 +
1 r r
z
z
( fz )
1 ( rf r ) 1 ( f ) ( f z )
+
+
2
r
z
z
r r
( f ) ( f z )
1 2 It
+
=
+
1 + z

z
1 zz
r
r r r
2
2
1 2 It
1 zz zz
+
+
+
=
2
2
2
2
1 + z
1
z
r
1 z
r
r r r
2
2
1 z z
+
+
2
2
z 2
r
2
2
( f z ) ( fr )
1 2 It
1 rz rz
+
+
+
=
+
2
2
2
z
1 + r z
z
r
r
( f r ) ( f )
1 r 1 2 r 2 r
1 2 It
r
+ 2
+
+
=
+
2
2
r r r r
z
r
1 + r

1 rz
r
r r r
It = rr + + zz
(1-68)
- 26 -
Beltrami-Michell Equations in Spherical Coordinates:

1 2 rr
r
r
r 2 r

rr
1
+ 2
sin
r sin
2 rr
1
1 2 It
+
+
=
2 2
2
1 + r 2
r sin
2
( fr )
1 ( r fr )
1
1 ( f )
2
2
+
( sin f ) +
1 r
r
r sin
r sin
r

1 2 It
1 2
1
1

sin
=
r
+
+
+
r 2 r
r r 2 sin
r 2 sin 2 2 1 + 2
2
( f )
1
1 ( f )
1 ( r fr )
2
2
+
( sin f ) +
r
r sin
r sin
1 r

2

1 2
1
1
1 2 It
+
+
+
=
r
sin
r r 2 sin
r 2 sin 2 2 1 + 2
r 2 r
2
2
( f )
1
1 ( f )
1 ( r fr )
2
2
+
( sin f ) +
r
r sin
r sin
1 r
1 2
r
r
r 2 r
1 2 r
r
r
r 2 r
1 2 r
r
r 2 r
r
( f ) ( f )

2

1
1
1 2 It
+
+
+
=
+
sin
2 2
2

r sin

1 +
r sin
2
2
( f r ) ( f )
r
r

1
1
1 It
+
=
+
+ 2
sin
+ 2 2
2

r sin
1 + r
r
r sin
( f r ) ( f )
r
2 r
1
1
1 2 It

+
=
+
+ 2
sin
+ 2 2
2
1 + r
r sin
r
r sin

It = rr + +
(1-69)
- 27 -
Chapter 2: Some Representative Boundary Value

Problems
Deformation of a Rod Standing in a Gravitational Field
As a first example in solving elasticity boundary-value problems, we consider a long rod
that stands freely in a gravitational field (see Fig. 2.1).
X
Fig. 2.1
We start with the stress boundary conditions.
Ti = ij n j
(2-1)
On the top surface we have:

Ti (0, 0, 0), ni (0, 0,1)
T1 = 11n1 + 12 n2 + 13 n3 0 = 0 + 0 + 13 13 = 0
T2 = 21n1 + 22 n2 + 23 n3 0 = 0 + 0 + 23 23 = 0
T3 = 31n1 + 32 n2 + 33 n3 0 = 0 + 0 + 33 33 = 0
On the lateral surfaces we have:
(2-2)
- 28 -
Ti (0, 0, 0), ni (n1 , n2 , 0)

0 = 11n1 + 12 n2 + 0
(2-3)
0 = 21n1 + 22 n2 + 0
0 = 31n1 + 32 n2 + 0
On the bottom surface we have:
Ti (0, 0, gL), ni (0, 0, 1)
0 = 0 + 0 13
(2-4)
0 = 0 + 0 23
gL = 0 + 0 33 33 = gL
We now turn our attention to the overall equilibrium:
ij , j + fi = 0
11,1 + 12,2 + 13,3 + f1 = 0
(i )
21,1 + 22,2 + 23,3 + f 2 = 0
(ii )
31,1 + 32,2 + 33,3 g = 0
(iii )
(2-5)
Near the bottom surface we have:
13 = 0, 23 = 0,
Thus, using (2-5 iii), we can write:
33,3 g = 0 33 = gz + f (1, 2)
B.C. : at z = 0 we have 33 = gL f (1, 2) = gL
(2-6)
33 = gz gL = g ( z L)
We may therefore assume that the distribution of the stress tensor has the following form:
0
0 0
0
ij = 0 0
0 0 g (L z)
(2-7)
- 29 -
Note that this form satisfies all the equilibrium equations and the boundary conditions of
the problem. Using the following constitutive equations:
ij =
(2-8)
1
(1 + ) ij ij kk
E
we can obtain the distribution of the strain tensor throughout the rod.
g
0
0
E (L z)
ij =
0
(L z)
0
0
0
( L z )
E
(2-9)
Since the strain components are linear in Z, the compatibility equations are all satisfied.
Now we use the kinematic equations to obtain the displacement field:
ij =
(2-10)
1
( ui, j + u j ,i )
2
We start with the normal strain components:
1
g
(L z)
u1,1 + u1,1 ) = u1,1 u1,1 =
(
2
E
g
( L z ) x + f1 ( y, z )
u1 =
E
1
g
22 = ( u2,2 + u2,2 ) = u2,2 u2,2 =
(L z)
2
E
g
( L z ) y + f 2 ( x, z )
u2 =
E
1
g
33 = ( u3,3 + u3,3 ) = u3,3 u3,3 =
(L z)
2
E
g
1 2
u3 =
Lz z + f 3 ( x, y )
2
E
11 =
(i)
(ii)
(iii)
Now we may invoke the displacement boundary conditions:
(2-11)
- 30 -
at x = 0, u1 = 0, f1 = 0
at y = 0, u2 = 0, f 2 = 0
In order to find f3, we calculate the shear strain components and use (i, iii) of (2-11) to
write:
1
( u1,3 + u3,1 ) u1,3 + u3,1 = 0
2
g
g
g 2
x + f 3,1 = 0 f 3,1 =
x f3 =
x + h1 ( y )
E
E
2E
13 =
(2-12)
Now, using the expressions (ii, iii) of Eqs. (2-11) along with (2-12), we may write:
1
( u2,3 + u3,2 ) u2,3 + u3,2 = 0
2
g
g
g 2
y + h1,2 = 0 h1,2 =
y h1 =
y +C
E
E
2E
23 =
(2-13)
Hence, the final form of the displacement equations can be written as:
u1 =
u2 =
g
E
g
E
(L z) x
(L z) y
1 2 g 2 g 2
x +
y +C
Lz z +
E
2 2E
2E
at x = 0, y = 0, z = 0, we have u = 0 C = 0
u3 =
u3 =
1 2 g 2 g 2
x +
y
Lz z +
E
2 2E
2E
(2-14)
- 31 -
Rotating Discs
In this section we study the stresses induced in rotating discs. Initially, we assume that
the thickness of the disc is small in comparison with its radius, so that the variation of
radial and circumferential stresses over the thickness can be neglected. We also assume
that zz = 0 .
Fig. 2.2
We may start by writing the equations of motion in cylindrical coordinates:
rr 1 r rz rr
+
+
+
+ f r = u&&r
r
r
r
z
r 1 z 2 r
+
+
+
+ f = u&&
r
r
r
z
rz 1 z zz rz
+
+
+
+ f z = u&&z
r
r
r
z
(2-15)
Considering our Basic assumptions, and due to the symmetry of geometry and loading,
we are left with only one equation:
rr rr
+
+ r 2 = 0
r
r
(2-16)
which can be rearranged into the following form:

d (r rr )
+ r 2 2 = 0
dr
(2-17)
Now we may set:

r rr = F (r )
dF (r )
+ r 2 2
dr
(2-18)
- 32 -
in which F(r) is a stress function.

The linear strain tensor in cylindrical coordinates is:
ur
rr = r
ij =
1 ur u u
+
r
r
r
1 u ur
=
+
r
r
2 r =
ur u z
+
z
r
u 1 u z
+
2 z =
z r
u
zz = z
z
2 rz =
(2-19)
which in this case reduces to:

ur
rr = r
ij =
r = 0
u
= r
r
rz = 0
z = 0
zz = 0
(2-20)
We may also write the constitutive equations for our problem as:
1
(1 + ) ij ij kk
E
1
1
rr = [ (1 + ) rr ( rr + ) ] = [ rr ]
E
E
1
= [ rr ]
E
ij =
(2-21)
At this point, we will eliminate the displacements between our kinematic equations to
obtain a compatibility equation.
d
d u
= r
dr
dr r
dur ur 1 dur ur 1
2 =
= [ rr ]
=
r dr
r r
rdr r
(2-22)
Now we substitute the constitutive equation in the above compatibility equation:

d 1
11
( rr ) = ( rr + rr )
dr E
r E
d
d
1
rr = (1 )( rr )
dr
dr
r
(2-23)
- 33 -
Rewriting the above equation in terms of our stress function F(r), we will have:
dF (r )
+ r 2 2 , F (r ) F
dr
d 2F
dF F (1 + ) F dF
2
2
r
+
+ 2 =
2 r 2

2
2
dr
r dr
r
r r dr
r rr = F (r ),
(2-24)
d 2F
dF
+r
F = (3 + ) 2 r 3
r
2
dr
dr
2
This is a nonhomogeneous, second order linear, ordinary differential equation with

variable coefficients and has the following general solution:
1 (3 + ) 2 r 3
F (r ) = A1r + A2
r
8
(2-25)
Accordingly, the resulting stresses will be:

A2 (3 + ) 2 r 2
8
r2
A (1 + 3 ) 2 r 2
= A1 22
8
r
rr = A1 +

(2-26)
The coefficients A1 and A2 can be found by imposing the appropriate boundary

conditions. We consider the solution for the Solid Disc, and a Disc with stress-free
circular hole.
For the solid disc we must have A2=0 to account for the finite stresses at the center. The
second B.C. is:
Ti = ij n j
Ti (0, 0, 0), ni (1, 0, 0) at r = a
Tr = rr nr + r n + rz nz 0 = rr nr + 0 + 0 rr = 0
A1
3 + ) 2 a 2
(
=
8
Accordingly, the final form of the stresses will be:
(2-27)
- 34 -
(3 + ) 2 2 2
rr =
(a r )
8
(3 + ) 2 2 (1 + 3 ) 2 2
=
a
r
8
8
(2-28)
The displacement field can be obtained as follows:

ur = r =
=
r
( rr )
E
r 2
( 3 2 2 ) a 2 + ( 2 1) r 2
8E
(2-29)
For a disc with stress-free circular hole, we have the following conditions at the outside
radius:
Ti = ij n j
Ti (0, 0, 0), ni (1, 0, 0) at r = a
(2-30)
2 2
A2 ( 3 + ) a
A1 = 2 +
a
8
At the inside radius we have:
Ti = ij n j
Ti (0, 0, 0), ni (1, 0, 0) at r = b
(2-31)
2 2
A2 ( 3 + ) b
A1 = 2 +
b
8
Solving (2-30) and (2-31) for A1 and A2 , we will have:

A1
3 + ) 2
(
=
(a
+ b2 )
8
( 3 + ) 2 a 2b 2
A2 =
8
(2-32)
- 35 -
Hence the stress equations become:
rr =

(3 + ) 2 2
a 2b 2
2
a
b
+
r2
2
8
r
(3 + ) 2 2
a 2b 2 (1 + 3 )r 2
2
a
b
=
+
+
8
r2
(3 + )
(2-33)
Furthermore, we may write the maximum radial and hoop stresses as:
rr

(3 + ) 2
2
(a b)
8
(3 + ) 2 2 2(1 )b 2
=
2a +
8
(3 + )
max
max
(2-34)
Also notice that for a small hole as b 0 , we will have:

hole
b0
r 0
solid
(3 + ) 2
=
2a 2 )
(
8
(3 + ) 2 2
a
=
8
hole
b 0
= 2
(2-35)
solid
r 0
The conclusion is that a stress concentration factor of two exists for the circular hole in
our rotating disc.
Alternative solutions
There are alternative solutions to the rotating disc problem, which are in fact quite
simpler than the solution presented above. For instance, we may extract our previous
differential equation in terms of the stress function F(r) directly from the BeltramiMichell equations in cylindrical coordinates (see Chapter 1).
We may also start with the Navier displacement equations in cylindrical coordinates:
( + 2G )
I e
1 z
2G
r
z
r
+ fr = 0
I e

2G r z + f = 0
r
r
z
I
(r ) r
( + 2G ) e 2G
+ fz = 0
z

r
( + 2G )
(2-36)
- 36 -
In the above equations we have:

1 (rur ) 1 u u z
+
+
r r
r
z
u u
1 u z u
2r =
; 2 = r z
r
z
z
r
I e = eii = u =
1 (ru ) ur
; 2z =
r r

(2-37)
Considering our basic assumptions and the symmetry of geometry and loading, the above
equations will simplify to the following single equation:
r2
d 2ur
dur
1 2
+
2 r 3
r
u
2
dr
dr
E
(2-38)
The general solution can be shown to be:

ur =
A2 1 2 2 2 3
1
1
1
A
r
r r
(
)
(
)
1
8
E
r
(2-39)
which upon substitution into the kinematic and constitutive equations will give the
required expressions for the radial and hoop stresses.
Assignment 4:
Find the stresses in a rotating disc of variable thickness h=f(r).
Hint: use the following form of the equilibrium equation:
d
( hr rr ) h + 2 hr 2 = 0
dr
- 37 -
Saint-Venant Torsion
This section presents a general solution to the torsion of prismatic bars, known as SaintVenant torsion problem. This solution represents a classic example of the so called semiinverse approach, in which a displacement field is initially assumed and then it is shown
that it satisfies the equilibrium equations and the boundary conditions.
X2
X2
R
C
X1
u
r u2 u1
X3
X1
Fig. 2.3
Based on a geometric study of the schematic shown in Fig. 2.3, we may define the two
in-plane components of displacement vector for an arbitrary point in the cross section of
the bar by:
u1 = r cos( + ) r cos
u2 = r sin( + ) r sin
u1 = r cos cos r sin sin r cos
u2 = r sin cos + r cos sin r sin
(2-40)
For very small values of , we may assume that cos 1 , sin , and rewrite the
displacements as:
u1 = r cos r sin r cos
u2 = r cos + r sin r sin
u1 = r sin
u2 = r cos
which using the expressions r cos = x1 , rsin = x2 , results in,
(2-41)
- 38 -
u ~ ( x2 , x1 )
(2-42)
With the definition of as the torsion-angle per unit length, we have:
= x3
(2-43)
Hence, the in-plane displacements can be written as follows:

u ~ (x2 x3 , x1 x3 )
(2-44)
Next, we assume that the third component of displacement can be defined as:
u3 = ( x1 , x2 )
(2-45)
where is called the Warping Function. Finally, the assumed displacement field can be
summarized as:
ui ~ (x2 x3 , x1 x3 , )
(2-46)
Accordingly, the strain field becomes:
1
( ui, j + u j ,i )
2
1
0
0
x2 + ,1 )
(
ij ~ 0
x1 + ,2 )
(
ij =
(2-47)
and the stresses become:
ij = 2 ij + ij kk
0 0 G (,1 x2 )
ij ~ 0 G (,2 + x1 )
Now we turn our attention to the equilibrium equations:
(2-48)
- 39 -
11,1 + 12,2 + 13,3 = 0

21,1 + 22,2 + 23,3 = 0
(2-49)
31,1 + 32,2 + 33,3 = G (,11 + ,22 )

which can be summarized as:
2 = 0 in R
(2-50)
Now we will examine the boundary conditions.

T = i ni
T (0, 0), ni (n1 , n2 , 0)
0 = 11n1 + 12 n2 + 0
0 = 21n1 + 22 n2 + 0
(2-51)
T3 = 3i ni = 0
0 = 31n1 + 32 n2 + 0
= G (,1 x2 ) n1 + (,2 + x1 ) n2 = 0
,1n1 + ,2 n2 = x2 n1 x1n2
We may simplify the above boundary conditions based on the geometric considerations
depicted in Fig. 2.4 as follows:
X2
n
t
X1
Fig. 2.4
- 40 -
dx1
dx
= cos , n2 = 2 = sin
dn
dn
dx
dx
n1 = 2 = cos , n2 = 1 = sin
dt
dt
n1 =
(2-52)
Accordingly, the stress boundary condition defined in (2-51) may be rewritten as:
dx1 dx2
dx
dx
+
= x2 2 + x1 1
dt
dt
x1 dn x2 dn
d 1 dr 2
=
dn 2 dt
(2-53)
on C.
Note that the above is not yet in a suitable form because we have to deal with two
variables, n and t. In order to solve this problem, we introduce the conjugate harmonic
function :
z = + i
(2-54)
Using the following Cauchy-Riemann conditions:
,1 = ,2
,2 = ,1
(2-55)
we will have:
2 = 0 in R
(2-56)
and the boundary conditions will be:

dx2 dx1 1 dr 2
+

=
x2 dt x1 dt 2 dt
d 1 dr 2
on C.
=
dt 2 dt
(2-57)
Thus, for a simply connected region we have:

1
2
= r 2 on C.
and the stresses become:
(2-58)
- 41 -
13 = G ( ,2 x2 )
23 = G ( ,1 + x1 )
(2-59)
We may further simplify our boundary value problem by another variable change, as
follows:
1
= r2
2
= 0 on C, and 2 = 2 in R.
(2-60)
In fact we may solve the torsion problem for different geometries through finding the
appropriate function that satisfies the conditions presented in (2-60).
Accordingly, the stresses become:
0 0 G ,2
ij ~ 0 G ,1

(2-61)
Circular Cross Section

For a circular cross section, we choose:
= k ( x12 + x22 a 2 )
(2-62)
which satisfies Eqs. (2-60). Next, we have:
= 0 at r = a
2 = 4 k
1
1
1
k = = ( x12 + x22 ) + a 2
2
2
2
(2-63)
so,
1
1
2
2
= constant = C
= + ( x12 + x22 ) = a 2 = constant
which means that the circular cross section does not warp in torsion.
(2-64)
- 42 -
Accordingly, the displacements and stresses are:

ui ~ (x2 x3 , x1 x3 , C )
0 0 Gx2
ij ~ 0 Gx1

0
(2-65)
Now we may calculate the total torque required to produce a certain amount of twist.
X2
32
dA
31
X2
X1
X1
Fig. 2.5
Bases on geometric considerations depicted in Fig. 2.5, we can write,
M = ( x1 32 x2 31 )dA
A
= G ( x12 + x22 ) dA
A
= G
r 2 rd dr
r4
= G
4 0
a4
= G 2
4
Ga 4
=
2
2
0
(2-66)
- 43 -
Elliptical Cross Section

For an elliptical cross section, we choose:
x2 x2
= k 12 + 22 1
a b
(2-67)
which satisfies Eqs. (2-60). Next, we have,

= 0 on C
a 2b 2
1 1
2 = 2 = 2k 2 + 2 k = 2
a b
( a + b2 )
=
(2-68)
ab x
x
( a 2 + b2 ) a b
2 2
2
1
2
2
2
2
Thus the stresses become:
2Ga 2
x2
0
0
2
2
+
b
a
2Gb 2
ij ~ 0
x
2
2 1
b +a
(2-69)
We also have,
1
2
2 2
a b x12 x22 1 2
1 2 2 + ( x1 + x22 )
= 2
2
(a + b ) a b 2
= + ( x12 + x22 )
(2-70)
a 2 b2
b2 a 2
a 2b 2
2
x22
+
+
x
1
2
2
2
2
2
2
( a + b ) 2 ( a + b ) 2 ( a + b )
Accordingly, we may calculate the warping function using the Cauchy-Riemann

conditions:
- 44 -
b2 a 2
x2
a 2 + b2
b2 a 2
= 2
x1 x2 + f ( x2 )
a + b2
,2 = ,1
,1 = ,2 =
(2-71)
b2 a 2
b2 a 2
2
x1 + f ,2 = 2
x1
a + b2
a + b2
f ,2 = 0 f = C
=
b2 a 2
x1 x2 + C
a 2 + b2
Accordingly, the out-of-plane displacement is:

b2 a 2
u3 = = 2
x1 x2
a + b2
(2-72)
Fig. 2.6 shows how an elliptical cross section warps due to the existence of the out-ofplane displacements.
x2
Fig. 2.6
Assignment 5:
Solve the torsion problem for a quadrilateral cross section.
x1
- 45 -
Cantilever Beam under End Loading

We start this section by verification of the basic equations governing the Cantilever Beam
from an elementary point of view, i.e., the Bernoulli-Euler beam theory. Later we will
present the theory of elasticity point of view. The Bernoulli-Euler beam theory is built
upon the following assumptions:
x1
2b
2h
x2
x3
Fig. 2.7
h << l , b << l
33 << 11 33 0
(2-73)
It is further assumed that Linear Elements which are initially normal to the undeformed
middle surface remain straight and normal to the deformed middle surface and suffer no
extension, so we may write:
Undeformed Geometry
x1
x3
Deformed Geometry
Fig. 2.8
11 12
ij ~ 22

- 46 -
0
23
0
(2-74)
On the other hand, the stresses are assumed to be:
13
11 0
0
ij ~ 0

33 0
(2-75)
The resultant moment is:

M = 11 x3 dA
(2-76)
and the resultant shear stress will be:

V = 13 dA
(2-77)
In order to verify the validity of the basic assumptions and the proposed stress and
strain fields, we start from the assumed Displacement field:
u1 = x3 ( x1 )
(2-78)
u2 = u2 ( xi )
= u3 = u3 ( xi )
in which is the angle of rotation before and after deformation. The strains can be
obtained using the kinematic equations as follows:
1
( ui, j + u j ,i )
2
u1,1 = x3 ,1
ij ~
ij =
1
u2,1
2
u2,2
+ ,1 )
(
2
11 12
u2,3 + ,2 ) ij ~ 22
(
,3
0
23
0
which, when compared to the assumed strain field, gives the following results:
(2-79)
- 47 -
13 = 0 = ,1
(2-80)
33 = 0 = ( x )
Now we obtain the strain field from the initially assumed stresses using the following
constitutive equations:
ij =
1
(1 + ) ij ij kk
E
11
ij ~

11
E
0
13
0
0 ij ~
33 0
11
(1 + ) 13
11
E
(2-81)
If we compare the resulting strains with the initially assumed strains and those obtained
directly from the displacements, we will notice how the troublesome terms in the strain
field have been ignored to match the stress and strain fields and obtain a workable theory.
u2,1 = 0
11
E = x3 ,1 = x3,11 u2 = u2 ( x2 , x3 )
ij ~
11 = u2,2
1
(1 + ) 13
( + ,1 ) =
is ignored
( u2,3 + ,2 ) = 0
u2,3 = ,2
E 11 = 0
( is ignored)
11
E
(2-82)
Next, we proceed with the derivation of other equations, and write:

u2,2 =
11 , 11 = Ex3,11
u2 = ,11 x2 x3
We also have:
(2-83)
- 48 -
1
( u2,3 + ,2 ) = 0
2
,2 = u2,3 = ,11 x2
23 =
(2-84)
and,
M = 11 x3 dA = E ,11 x32 dA = E,11 I
A
M
,11 =
EI
(2-85)
Hence, we may rewrite the displacements as:

u2 =
M
EI
x2 x3 ; ,2 =
M
EI
x2
(2-86)
Now we check the Equilibrium Equations:
11,1 + 12,2 + 13,3 = 0 11,1 + 13,3 = 0

21,1 + 22,2 + 23,3 = 0 identically satisfied
(2-87)
31,1 + 32,2 + 33,3 = 0 31,1 + 33,3 = 0

Multiplying the first equilibrium equation by x3 and integrating across the cross section,
we will have:
b
h
x3 11,1dA + x3 13,3 dx3 dx2 = 0
h
A
b
( x dA) +
x1
b
M
h
+ x3 13 h dx2 V = 0
b
x1
x h h dx dx = 0
3 13 h h 13 3 2
b
11
(2-88)
For problems with no surface shear force:

M
V = 0
x1
Integrating the third equilibrium equation across the cross section, we will have:
(2-89)
- 49 -
b h
13
33
dA +
dx3 dx2 = 0
A x
b
h x
1
3
x1
( dA) +
A
13
(2-90)
33 33 dx2 = 0
h
h
Now we define the resultant normal surface load as:

b
q = 33 h 33 h dx2
b
(2-91)
and the Eq. (2-90) becomes:

V
+q =0
x1
(2-92)
We may also derive other important equations for the B.E. beam as follows:
M
V = 0
x1
2 M V
2M
=
=
+q=0
x12
x1
x12
,11 =
M
EI
1 2M
q
4
=
=
4
2
EI x1
EI
x1
(2-93)
q
4
=
4
EI
x1
Finally, the following equations are considered to form the basis of the B.E. Beam theory.
,11 =
M
,
EI
M
V = 0,
x1
V
+ q = 0,
x1
2M
+ q = 0,
x12
4 q
=
,
x14 EI
(2-94)
- 50 -
Solution Based on the Theory of Elasticity

We initially assume that the normal stresses are the same as those of the elementary
theory and also 32 is taken as zero, but no stipulation is made regarding the other shear
stresses 13 and 12. Accordingly, the three equilibrium equations become:
11,1 + 12,2 + 13,3 = 0 12,2 + 13,3 =

21,1 + 22,2 + 23,3 = 0 21,1 = 0
Px3
I
31,1 + 32,2 + 33,3 = 0 31,1 = 0
(i)
(2-95)
(ii)
From equations (2-95, i) and (2-95, ii), we may conclude that the shearing stresses do not
depend on x1, so they are the same for all cross sections. For this case the two relevant
stress compatibility equations become:
mm,ij = fi , j + f j ,i +
ij f m,m
1 +
1
1 P
2 31 =
(i )
1 + I
2 21 = 0
(ii )
2 ij +
(2-96)
Now we must ensure that the modified solution satisfies the boundary conditions.
31 ( h, x2 , x1 ) = 0
23 ( x3 , b, x1 ) = 0
(2-97)
At this stage we define our stress components in terms of a stress function:
31 = ,2
21 = ,3
Px32
+ f ( x2 )
2I
(2-98)
which satisfies the equilibrium equations. Then our first stress compatibility equation
(Eq. (2-96, i)) becomes:
P
1 P
(,233 + ,222 ) + f ,22 =
1 + I
I
(2-99)
P
(,33 + ,22 ),2 =
f,22
1 + I
P
,33 + ,22 =
x2 f,2 + C
1 + I
- 51 -
And the second compatibility equation (Eq. (2-96, ii)) is:
,333 ,322 = 0
(,33 + ,22 ),3 = 0
(2-100)
Note that equation (2-99) also satisfies equation (2-100). In order to find C, we express
the rotation about the x1 axis by:
32 =
1
( u3,2 u2,3 )
2
(2-101)
and its rate of change by:
1
( u3,2 u2,3 ),1
2
1
= ( u3,1 + u1,3 ),2 ( u2,1 + u1,2 ),3
2
= 31,2 21,3
32,1 =
(2-102)
which can be written in terms of stresses using the constitutive equations.

1
( 31,2 21,3 )
2G
1
=
(,22 + ,33 + f,2 )
2G
32,1 =
(2-103)
Also we can combine (2-103) and (2-99) to write:

2G32,1 =
1 P
x2 + C
1 + I
(2-104)
As we are considering a rectangular cross section and the bending is symmetrical about
the x3 axis, at x2=0 we should have the average rotation equal to zero, resulting in C=0.
Hence, equation (2-99) becomes:
,33 + ,22 =
P
x2 f ,2
1 + I
(2-105)
which is the governing equation for the stress function, provided that the resulting
stresses satisfy the boundary conditions.
- 52 -
However, we may write the boundary conditions directly in terms of the stress function
by choosing a proper form for the function f(x2), as follows:
Ph 2
2I
31 ( h, x2 , x1 ) = ,2 ( h, x2 , x1 ) = 0
23 ( x3 , b, x1 ) = ,3 ( x3 , b, x1 ) = 0
f ( x2 ) =
(2-106)
and the governing equation changes to:
,33 + ,22 =
P
x2
1 + I
(2-107)
Hence, in order to solve the problem at hand, we should choose a proper form of such
that it satisfies the above equation and vanishes on the boundary.
Membrane Analogy
At this stage we may draw an analogy between the above governing equation and the
equation of a homogeneous membrane whose outline is the same as the cross section of
our beam. First, let us start with studying the deformation of a membrane subjected to a
uniform tension at the edges and a uniform lateral pressure (see Fig. 2.9).
X1
S
X3
X1
dx
dy
X2
Fig. 2.9
- 53 -
If q is the pressure per unit area of the membrane and S is the uniform tension per unit
length of its boundary, the tensile forces acting on the opposite sides (dy sides) of an
infinitesimal element give a resultant in the upward direction: S ( 2 z / x 2 )dxdy . In the
same manner, the tensile forces acting on the other two opposite sides (dx sides) of the
element give the resultant S ( 2 z / y 2 )dxdy (see Section 11.7, Kreyszig, Advanced
Engineering Mathematics) and the equation of equilibrium of the element becomes:
qdxdy + S
2 z
2 z
dxdy
S
dxdy = 0
+
x 2
y 2
(2-108)
q
2 z 2 z
2 + 2 =
S
x y
At the boundary, the deflection of the membrane is zero. Comparing the above equation
with the governing equation of our beam problem, it is clear that the solution of the beam
problem reduces to the determination of the deflections of the membrane which are
produced by a continuous load with the intensity proportional to / (1 + )( Px2 / I ) .
The curve mnp in Fig. 2.10 represents the intersection of the membrane with the x1x2
plane.
X2
b
X3
X2
X1
Fig. 2.10
- 54 -
Now we rewrite the stresses from equation (2-98) (incorporating f ( x2 ) = Ph 2 / 2 I ):

Px32 Ph 2
+
2I
2I
2
2
P(h x3 )
=
+ ,2
2I
21 = ,3
31 = ,2
(2-109)
From the above equations it is clear that the deviations of the shearing stresses from the
elementary theory are proportional to the first derivatives or slope of the membrane at
any point. For example, the correction to 31 shows a maximum positive value at the two
sides, a maximum negative value in the center, and is zero at the quarter points (See Fig.
2.10). From the condition of loading of the membrane it is clear that is an even
function of x3 and an odd function of x2. Hence, it is appropriate to take the stress
function in the form of the Fourier series:
m = n =
m = 0 n =1
2 m +1, n
cos
( 2m + 1) x3 sin n x2
2h
(2-110)
Substituting from above expression into our governing equation we may find the
coefficient of the Fourier series and obtain the final form of our stress function as:
P 8b3
=
1 + I 4
m = n =
m = 0 n =1
( 1)
m + n +1
cos
( 2m + 1) x3 sin n x2
2h
2 b
( 2m + 1) n ( 2m + 1) 2 + n 2
4h
(2-111)
which can be used to calculate the shearing stresses using equation (2-98).
Note that the above membrane analogy and Eq. (2-108) can also be used to obtain
solutions to the TORSION problems discussed before.
- 55 -
Approximate Solutions
If the depth of the beam is large compared with the width h>>b, we may consider a
cylindrical shape for the surface of the membrane at the points sufficiently far from the
short sides at x3 = h (assume = ( x2 ) ), so we may write:
= ( x2 )
Px2
1 + I
P 3
( x2 ) =
( x2 + C1 x2 + C2 )
1 + 6 I
(0) = 0 C2 = 0
2
(b) = 0 C1 = b
P 3 2
( x2 ) =
( x2 b x2 )
1 + 6 I
P 2
2
31 ( x3 , x2 ) =
( h x3 ) +
2I
1 +
,22 =
(2-112)
2 b2
x2
3
On the other hand, if the width of the beam is large compared to the depth (b>>h), for
the points far from the short sides at x2 = b , we may consider the deflections of the
membrane as a linear function of x2 and write:
Px2
1 + I
Px2 x3
,3 =
+ C1
1 + I
Px2 2
x3 + C1 x3 + C2
( x3 , x2 ) =
1 + 2 I
Px2 h 2
( h, x2 ) = 0 C2 =
, C1 = 0
1 + 2 I
Px2 2 2
( x3 , x2 ) =
( x3 h )
1 + 2 I
2
2
P ( h 2 x32 )

1 P ( h x3 )
31 =
1 1 + = 1 +
2I
2I
,33 =
21 =
Px3 x2
1 + 2 I
(2-113)
- 56 -
Anti-Plane Shear Problem

Consider a cylindrical solid with arbitrary cross-section, as shown in the Fig. 2.11. Also
assume that the length of the cylinder greatly exceeds any cross sectional dimension. We
may have states of anti-plane shear in a solid by an appropriate loading, as we will see in
the following example.
The governing equations and boundary conditions for anti-plane shear problems are very
simple and lead to solving the Laplaces equation (for which we have many powerful
solution techniques).
We consider the following boundary value problem, with body force b = b3 ( x1 , x2 )e 3
T1 = T2 = 0
on
u3 = u ( x1 , x2 )
on
Cu
T3 = T3 ( x1 , x2 )
on
CT
(2-114)
X2
R1
X1
R2
Anti Plane Shear
X3
C
Fig. 2.11
In addition, for a traction boundary value problem we must ensure the existence of a
static equilibrium solution:
( x1 , x2 )ds + b3 ( x1 , x2 )dV = 0
R
(2-115)
- 57 -
To accept any solution with zero resultant force and moment acting on the ends of the
cylinder, we set the boundary conditions on R ( = 1, 2) as:
n dA = 0
(2-116)
r n dA = 0
Recall the field equations:
ij = 12 (ui , j +u j ,i )
(2-117)
ij = ij kk + 2 ij
ij , j +bi = 0
Since all forces and boundary displacements act in the x3 direction, it is natural to assume
that the displacements are in the x3 direction everywhere. Lets assume a solution of the
form:
u1 = u2 = 0
u3 = u%3 ( x1 , x2 )
(2-118)
The strains and stresses follow as:
= 0,
33 = 0
3 = 12 u%3 ,
= 0,
33 = 0,
3 = u%3 ,
(2-119)
where Greek subscripts range from 1 to 2. The equilibrium equations reduce to:
3 , +b3 = 0
u%3 , +b3 = 0
u%3 , =
(2-120)
b3
The boundary conditions may be re-written as,

u%3 = u3 ( x1 , x2 )
on
T3 = 3 n = u%3 , n = T3 ( x1 , x2 )
Cu
on
CT
(2-121)
- 58 -
As mentioned before, the Laplaces equation can be solved in different ways. Here we
will use the complex variable method. This approach is based on the fact that both the
real and imaginary parts of an analytic function satisfy the Laplaces equation. We start
with the following definitions:
f ( z ) = v( x1 , x2 ) + iw( x1 , x2 )
z = x1 + ix2
i = 1
(2-122)
If f ( z ) is analytic, the derivative with respect to z is path independent, which requires:

v w
=
x1 x2
w
v
=
x1
x2
(2-123)
These are known as the Cauchy-Riemann conditions. Hence, we have:

v,
2v 2v
w w
= 2+ 2 =
=0
x1 x2 x1 x2 x2 x1
(2-124)
or, in general:
v, = w, = 0
(2-125)
Thus, both the real and imaginary parts of an analytic function satisfy the Laplace
equation.
Now, let z = x1 + ix2 = rei characterize the position of a point in the plane of the solid of
interest. Then let f ( z ) denote any analytic function of position. We can set:
u%3 = e( f ( z )) = v( x1 , x2 )
(2-126)
Note that u%3 automatically satisfies the equilibrium equation u%3, = 0 . We can solve our
elasticity problem by finding an analytic function that satisfies appropriate boundary
conditions. Of course, we may also use:
u%3 = Im( f ( z )) = w( x1 , x2 )
(2-127)
Now, we can determine the stresses directly from f(z). Note that
f ( z ) = v + iw
f ( z ) = v,1 +iw,1 = v,1 iv,2 = w,2 +iw,1
(2-128)
- 59 -
Suppose we choose u%3 = e( f ( z )) , then using (2-119) we have:

f ( z ) = u%3 ,1 iu%3 ,2
f ( z ) = 31 i 32
(2-129)
f ( z ) = 31 + i 32
Where f ( z ) denotes the complex conjugate.

Similar expressions can be determined for u% = m( f ( z )) . In this case
f ( z ) = u%3 ,2 +iu%3 ,1
f ( z ) = 32 + i 31
(2-130)
As an example we consider the fundamental solution for an infinite solid and find the
displacement and stress fields induced by a line load F = Fe 3 acting at the origin.
We start by generating the solution from the following function:
f ( z ) = C log( z )
(2-131)
X2
X1
Fig. 2.12
This is analytic everywhere except the origin. Next we choose:

u%3 = e( f ( z ))
(2-132)
- 60 -
and compute the resultant force acting on a circular arc enclosing the origin:
B
F3 = ( 31n1 + 32 n2 )ds =
A
31
cos + 23 sin )rd
(2-133)
Here we may write:
31 cos + 32 sin = e ( 31 + i 32 ) (cos i sin )

= e( f '( z )e i )
(2-134)
which using (2-131), can be written as:

e i
e i
= e C i = C
re
r
31 cos + 32 sin = e C
(2-135)
Substituting (2-135) into the right hand side of (2-133), results in:
F3 = 2 C
C =
F3
2
(2-136)
Therefore, the solution can be found and simplified as follows:

F3
e {log( z )}
2
F
F
u%3 = 3 e[log(rei )] = 3 log(r )
2
2
u%3 =
r = x x
Assignment 6:
Find the stress field for a Screw Dislocation problem.
b
Hint: Assume u% =
m {log( z )}
2
(2-137)
- 61 -
The Mode III Crack Problem

The analysis of Mode III crack problem is relatively simple because we may assume that
u1=u2=0 and u3 = u3(x1 , x2), which clearly represent an Anti-Plane Shear problem.
III
Fig. 2.13
Referring to the material covered in the above sections, and ignoring the body forces in
our crack problem, we may write:
u%3 , = 2u%3 = 0
(2-138)
Here, we consider a solution in the following form:

u%3 =
f ( z) + f ( z)
(2-139)
Hence, the strains become:

1
f ( z ) + f ( z )
2
i
f ( z ) f ( z )
32 =
31 =
(2-140)
Using the constitutive equations we find that:
31 i 32 = 2 f ( z )
(2-141)
Now, let the origin of the coordinate system be located at the tip of a crack lying along
the negative x1 axis as shown in Fig. 2.14.
- 62 -
X2
X1
Fig. 2.14
Now, we will focus our attention upon a small region (denoted as D) containing the crack
tip and consider the holomorphic function:
f ( z ) = Cz +1 , C = A + iB
(2-142)
where A, B, and are real undetermined constants. For finite displacements at the crack
tip we must have: ( z = r = 0), f 1 ). Then we will have:
31 i 32 = 2( + 1)Cz = 2( + 1)r ( A + iB)(cos + i sin )
(2-143)
where,
31 = 2( + 1)r ( A cos B sin )

32 = 2( + 1)r ( A sin + B cos )
(2-144)
The boundary condition that the crack surfaces be traction free requires that 32 = 0 on
= . Consequently we have:
A sin + B cos = 0
(2-145)
A sin B cos = 0
To avoid the trivial solution, the determinant of the coefficients of the above equations
must vanish. This leads to:
sin 2 = 0
(2-146)
which for > 1 has the following roots:

1
2
= , n / 2, n = 0,1, 2,...
(2-147)
- 63 -
Of the infinite set of functions of the form of the above equation that yield traction-free
crack surfaces within D, the function with = 1/2 for which A = 0, provides the most
significant contribution to the crack-tip fields. For this case the stresses and
displacements become, respectively:
31
K III
=
1/ 2
32 (2 r )
sin( / 2)
cos( / 2)
(2-148)
and,
1
2 K III r 2
u%3 =
sin( / 2)
2
(2-149)
where B has been chosen such that:

K III = lim {(2 r )1/ 2 32
r 0
=0
(2-150)
The quantity KIII is referred to as the Mode III stress intensity factor, which is
established by the far field boundary conditions and is a function of the applied
loading and the geometry of the cracked body. Whereas the stresses associated with
the other values of are finite at the crack tip, the stress components associated with =
1/2 have an inverse square root singularity at the crack tip. It is clear that the latter
components will dominate as the crack tip is approached and represent the asymptotic
forms of the elastic stress and displacement fields.
- 64 -
Chapter 3
2D Static Boundary Value Problems: Plane Elasticity
Chinese Proverb - It is better to ask a question and look like a fool for five minutes, than not
to ask a question at all and be a fool for the rest of your life.
Two Dimensional Elastostatic Problems

Plane Strain Formulation
Consider a long prismatic bar loaded with surface tractions and body forces normal to the
x3 axis as depicted in Fig 3.1.
X2
X1
f
T
f
T
Plane stress
R
Plane strain
X3
C
Fig. 3.1
Also assume that the magnitude of loadings does not vary along this axis. Under these
conditions we may write:
- 65 -
u = u ( x , x )
u = u ( x ) 1 1 1 2 , = 1, 2
u2 = u2 ( x1 , x2 )
u3 = constant, generalized plane strain
(3.1)
u3 = 0, plane strain
Accordingly, the strain field becomes:
11 12
1
ij = ( ui , j + u j ,i ) ~ 22
2

0
0
(3.2)
In order to calculate the stress field we use the following constitutive equations:
ij = kk ij + 2G ij
(3.3)
Partitioning the stress tensor, we have

ij ~
3
33
(3.4)
Here we have three constitutive equations:
= + 2G
(i)
3 = 3 = 0
33 = = ( 11 + 22 )
(ii)
(3.5)
Contracting on in equation (3.5i) we have:
= 2 + 2G
=
1

2 ( + G )
(3.6)
Using the above equation and substituting for in equation (3.5ii) we can write:
33 =
=
2( + G)
(3.7)
which shows that in plane strain condition the stress in the x3 direction is a dependent
quantity. Accordingly, the strains can be written as:
- 66 -
1

2G
1
=

2G
2 ( + G )
1

=
2G
(3.8)
It is important to note that the above equation cannot be obtained from the original
constitutive equation by simply replacing i,j by ,.
1
2G 11 ( 11 + 22 )
~
1
22 ( 11 + 22 )
2G
1
12
2G
(3.9)
The equilibrium equations in the plane strain case become:
, + f = 0
11,1 + 12,2 + f1 = 0
12,1 + 22,2 + f 2 = 0
(3.10)
Now, we assume that the body force f can be derivable from a potential, V. Hence, we
may write:
f = V,
, V, = 0
( 11 V ),1 + 12,2 = 0
( 22 V ),2 + 12,1 = 0
(3.11)
The above equations can be satisfied by assuming the existence of two functions A(x)
and B(x), such that:
11 V = A,2 ; 12 = A,1
22 V = B,1 ; 12 = B,2
B,2 + A,1 = 0
(3.12)
The latter is satisfied by assuming the existence of a stress function such that:
B = ,1 ; A = ,2
(3.13)
- 67 -
Accordingly, the stresses can be written in terms of the Airy stress function as:
11 V = ,22
(3.14)
22 V = ,11
12 = 21 = ,12
The stress boundary conditions for the plane strain reduce to:
T = n
(3.15)
From the six stress compatibility equations we have only one important non-trivial
equation, which upon substitution of the stresses by the stress function becomes:
1
2V
mm,33 =
1 +
1
2
2
), = 2 + 2
x1 x2
2 33 +
2 = (
2 =
2V
1
2V
1
1
2 (,22 + ,11 ) =
2 2V
1
1 2 2
22 =
V
1
2 (V + ,22 + V + ,11 ) =
1 2
4 =
1
(3.16)
2
V
The above is the governing partial differential equation for the Airy Stress Function. If
the potential function is harmonic, i.e., 2V = 0 , then satisfies the homogeneous
biharmonic equation:
4 = 0
4 = 2 2 = (
),
4
4
4
+
+
2
x14
x12 x22 x24
(3.17)
- 68 -
Plane Stress Formulation

Let us consider a short prismatic body loaded with surface tractions and body forces
normal to the x3 axis as depicted in Fig. 3.1. Under these conditions we may write for the
stresses:
ij =
(3.18)
1 h/2
ij dx3
h h / 2
The bar sign over a quantity indicates that it has been averaged over the thickness. The
stress tensor is:
11 12
ij ~ 22

0
0
(3.19)
Also, the equilibrium equations can be written as:

f = V,
, V, = 0
( 11 V ) + 12,2 = 0
,1
( 22 V ),2 + 12,1 = 0
(3.20)
which are exactly the same as those derived for plane strain condition, except for the
mean value interpretation. Thus we may similarly write:
11 V = ,22
22 V = ,11
(3.21)
12 = 21 = ,12
and the related constitutive equations are:
1 +
= +

3 = 0
33 =
E
From Eq. (3.22i) we have:
(i)
(3.22)
(ii)
- 69 -
1 +
1
= 2 +
=

E
E

1
(3.23)
(3.22ii) 33 =
which shows that in plane stress condition the strain in the x3 direction is a dependent
quantity. We may also write stresses in terms of strains:
E
1 +
+ E
(3.24)
Substituting for we will have:

E

1 +
2G
= 2G +

+ 2G
(3.25)
If we compare the above equations with those used in plane strain, we conclude that the
entire system of equations in plane stress are analogous to those for plane strain, provided
that we replace the barred quantities with the unbarred and define the followings:
G =G
(3.26)
2G
+ 2G
Hence, we can derive the fundamental biharmonic boundary value problem for the plane
stress conditions as:
1 2
4 =
1
2
V
1 +
Assignment 7:
Find the stress field for a narrow rectangular beam under end loading.
(3.27)
- 70 -
Airy Stress Functions in Cylindrical Coordinates

It is quite beneficial to solve the boundary value problems involving cylindrical regions
using appropriate cylindrical coordinates.
X2
rr
r
rr
X1
Fig. 3.2
In case of zero or constant body forces, the state of stress is related to the Airy function
using the following transformations:
rr = ri rj ij
= i j ij
(3.28)
r = ri j ij
which, by appropriate substitutions for the stress function , lead to the following
expressions for the stresses:
1 1 2 1
1
+ 2
= ,r + 2 ,
2
r r r
r
r
2
= 2 = ,rr
r
1
1 1
=
= 2 , ,r
r
r r r
rr =

r
(3.29)
The governing equations for this coordinate system will be:
1
1
2 = ,rr + ,r + 2 ,
r
r
2

1 1 2
1
1
4
= 2 +
+ 2
+ ,r + 2 , = 0
2 , rr
r r r
r
r
(3.30)
- 71 -
As an example, we will obtain the stress field due to a line load of magnitude P per unit
out-of-plane length, acting on the surface of a homogeneous isotropic half-space.
P
X1
X3
Fig. 3.3
Assume the Airy function in the form of:
(3.31)
r sin
The resulting state of stress can be written as:
rr =
2 P cos
r
= r = 0
(3.32)
In order to find the displacement field we need to determine the strains by substituting the
expression for stress into the constitutive relation and then integrating the strains to
calculate the displacements. For the point force solution, one we can show that:
ur =
P cos log r
1 2
P sin
2
1
u =
P sin log r +
P sin
Axisymmetric Stress Distribution

When we have symmetrical stress distribution about an axis, in the absence of body
forces, we may write:
(3.33)
- 72 -
1 1
= , r
r r r
2
= 2 = ,rr
r
=0
rr =

r
(3.34)
and also:
1
2 = ,rr + ,r
r
1 1

4 = r ( r,r ),r = 0
r r
,r ,r
(3.35)
The above expression can successively be integrated with respect to r to give the general
form of the stress function for axisymmetric stress distribution in cylindrical coordinates
as follows:
1

r ( r,r ),r = 0
,r ,r
r
1

r ( r,r ),r = C1
,r
r
C
1
( r,r ),r = 1
r
,r r
1
( r,r ),r = C1 ln r + C2
r
( r,r ),r = C1r ln r + C2 r

r2
r2
r2
r,r = C1 ln r + C2 + C3 = C1r 2 ln r + C2 r 2 + C3
4
2
2
C
,r = C1r ln r + C2 r + 3
r
2
2
r
r
r2
= C1 ln r + C2 + C3 ln r + C4
4
2
2
(3.36)
= C1r 2 ln r + C2 r 2 + C3 ln r + C4
Accordingly the stresses become:
1
r2
1
= C1 ( 3 + 2 ln r ) + 2C2 C3 2
r
=0
rr = C1 (1 + 2 ln r ) + 2C2 + C3

r
(3.37)
- 73 -
Finally, the displacements are:

1
1 +
+ C4 sin + C5 cos
C1r (1 )( 2 ln r 1) 2 + 2C2 (1 ) r C3
E
r
1
u = [ 4C1r + C4 cos C5 sin + C6 r ]
E
ur =
(3.38)
Axisymmetric Stress and Displacement fields
When we also have axisymmetric displacements in the absence of body forces, the
displacement values do not depend on , so we have C1=0 and we may write:
1
r2
1
= C2 C3 2
r
=0
rr = C2 + C3

r
(3.39)
Assignment 8:
Find the stress field for a cylindrical pressure vessel under internal and external pressures.
Curved Beam under constant bending moment

As an example for two-dimensional problems in cylindrical coordinates we consider the
solution of a curved beam with rectangular cross section, under a constant bending
moment M.
RO
r
Ri
Fig. 3.4
rr
- 74 -
For the upper and lower faces, we can write:
rr ( R0 ) = rr ( Ri ) = 0
r ( R0 ) = r ( Ri ) = 0
(3.40)
On the other hand, the equilibrium conditions dictate that:
dA = 0
r dA = M
A
A
(i)
(3.41)
(ii)
Notice that in this problem the stress field is axisymetric but the displacement field is not,
so the stresses can be written as:
1
r2
1
= C1 ( 3 + 2 ln r ) + 2C2 C3 2
r
=0
rr = C1 (1 + 2 ln r ) + 2C2 + C3

r
(3.42)
Next, we write the expression for rr at the boundaries:

C1 (1 + 2 ln R0 ) + 2C2 + C3
1
=0
R02
1
C1 (1 + 2 ln Ri ) + 2C2 + C3 2 = 0
Ri
(3.43)
Note that we still need a third equation to be able to determine our three constants. In
order to find it, we write equation (3.41i) in terms of the stress function as follows:
Ro
Ro
dA = B ,rr dr = B ,r R = 0
Ri
(3.44)
1
Since we have rr = ,r , the above expression is consistent with our first boundary
r
condition and indicates that ,r = 0 on the boundary.
Also, the second equilibrium condition (3.41ii) can be written as:
- 75 -
Ro
r dA = B r, rr dr = M
Ri
Ro
Ro
Ri
Ri
B r,rr dr = B r,r
( Ro ) ( Ri ) =
Ro
B ,r dr = M
(3.45)
Ri
M
B
Considering the general form of the stress function in cylindrical coordinates:
= C1r 2 ln r + C2 r 2 + C3 ln r + C4
(3.46)
we may write (3.45) as:

C1 ( Ro 2 ln Ro Ri 2 ln Ri ) + C2 ( Ro 2 Ri 2 ) + C3 ln
Ro
M
=
Ri
B
(3.47)
which gives the required third equation in terms of the three constants of the problem.
Accordingly, we will have:
2M
Ro 2 Ri 2 )
(
NB
M 2
C2 =
Ro Ri 2 ) + 2 ( Ro 2 ln Ro Ri 2 ln Ri )
(
NB
4 M 2 2 Ro
C3 =
Ro Ri ln
NB
Ri
C1 =
(3.48)
where
R
N = ( Ro Ri ) 4 Ro Ri ln o
Ri
2
Hence, the stresses can be found as:
rr =
R
r
4 M 2 2 Ro 1
+ Ro 2 ln + Ri 2 ln i
Ro Ri ln
2
NB
Ri r
Ri
r
R
r
4 M 2 2 Ro 1
Ro 2 ln
Ri 2 ln i Ro 2 + Ri 2
Ro Ri ln
2
NB
Ri r
Ro
r
r = 0
(3.49)
- 76 -
Sheet with a Circular Hole under Uniaxial Stress

As an example for two-dimensional problems in cylindrical coordinates we consider a
thin sheet, with a circular hole, under uniaxial uniform tension as follows:
11 = T1
22 = 12 = 0
(3.50)
The above uniform stress is disturbed by the presence of the circular hole. Also note that
it is more convenient to define the local stress distribution near the hole using cylindrical
coordinates. Hence, we use the St.Venants principle and argue that the stress distribution
near the hole remains unchanged if we construct a hypothetical circle, with a diameter
equal to the sheet width, and apply equivalent forces on the boundary of this circle, as
shown in Fig. 3.5a.
(b)
T1
(a)
T1/2
X2
= +
(T1/2)(cos2)
(c)
T1
-(T1/2)(sin2)
X1
Fig. 3.5
- 77 -
We start by a transformation of the Cartesian components of stress into the Polar

components as:
rr = ri rj ij = 11 cos 2 + 2 12 sin cos + 22 sin 2

= i j ij = 11 sin 2 2 12 sin cos + 22 cos 2
r = ri j ij = ( 22 11 ) sin cos + 12 ( cos sin )

2
(3.51)
which, in combination with the stress state expressed by equation (3.50), give the stresses
at the location of the hypothetical circle as follows:
T1
(1 + cos 2 )
2
T
( b, ) = T1 sin 2 = 1 (1 cos 2 )
2
T
r ( b, ) = T1 sin cos = 1 sin 2
2
rr ( b, ) = T1 cos 2 =
(3.52)
As depicted in Fig. 3.5, we may separate the stresses along the outer boundary into two
parts. For the first part (shown in Fig. 3.5b) we have the axisymmetric stress state:
T1
2
(1)
r ( b, ) = 0
rr(1) ( b, ) =
(3.53)
Note that does not act on the boundary. The second part is defined for the circle
depicted in Fig. 3.5c in the form of a -dependent stress state:
T1
cos 2
2
T1
r(2)
sin 2
( b, ) =
2
rr(2) ( b, ) =
(3.54)
The solution for the axisymmetric stress state (represented by Eqs. (3.53)) can be easily
obtained from the solution of a circular disc under radial load (see Eq. (3.39)) as follows:
1
r2
1
= C2 C3 2
r
=0
rr = C2 + C3

r
(3.55)
- 78 -
We may find the constants using the following boundary conditions:

T1
2
(1)
r ( b, ) = 0
rr(1) ( b, ) =
(3.56)
rr(1) ( a, ) = 0
r(1) ( a, ) = 0
Accordingly, the stresses can be written as;
T1 b 2
T1 a 2b 2
2 b2 a 2 2 b2 a 2
T b2
T a 2b 2
(1) ( r , ) = 1 2 2 + 1 2 2
2 b a
2 b a
rr(1) ( r , ) =
1
r2
1
r2
(3.57)
In order to find the solution to the second part, we start with the general form of the
biharmonic equation. Due to the periodicity of the boundary conditions we consider a
proper form for the stress function and proceed as follows:
2 1 1 2
1
1
4 = 2 +
+ 2
+ ,r + 2 , = 0
2 , rr
r r r
r
r
r
= F (r ) cos 2
2 1 1 2
1
4
cos 2 F,rr + F,r 2 F = 0

+ 2
2+
2
r r r
r
r
r
2
9
9
F,rrrr + F,rrr 2 F,rr + 3 F,r = 0
r
r
r
(3.58)
The above differential equation can be solved for F and will eventually result in the
following expression for our stress function:
= C1 + C2 r 2 + C3r 4 + C4
1
cos 2
r2
(3.59)
Equation (3.59) along with equations (3.29) can be used to calculate the stresses as
follows:
- 79 -
1
1
+ 2C2 + 6C4 4 cos 2
2
r
r
1
= 2C2 + 12C3 r 2 + 6C4 4 cos 2

r
1
1
= 2C1 2 + 2C2 + 6C3 r 2 6C4 4 sin 2

r
r
rr(2) = 4C1
(2)
r(2)
(3.60)
The boundary conditions are:

T1
cos 2
2
T1
r(2)
sin 2
( b, ) =
2
rr(2) ( a, ) = 0
rr(2) ( b, ) =
(3.61)
r(2)
( a, ) = 0
which can be rewritten as:
4
a2
2
a2
4
b2
2
2
b
6a 2
6b 2
6
0
a4
C
1
6
0
4
a C2 T1
=
6 C3 2
4
b C4 T1

6
2
4
b
(3.62)
The above four equations can be solved for the four unknown constants. However, we
may simplify Eqs.(3.62) for the case where the hole is small compared to the width of the
plate. First, we multiply the forth equation by (a/b)2 and let a / b 0 , which gives C3=0.
Next, we multiply the third equation by a2, let a / b 0 , and get C2=-T1/4. Hence, we
may write the first and the second equations as:
T 6
4
C1 + 1 4 C4 = 0
2
a
2 a
2
T 6
2 C1 1 4 C4 = 0
a
2 a
from which we find:
(3.63)
- 80 (3.64)
a2
a4
C1 = T1 and C4 = T1
2
4
Finally, we obtain the stresses using Eq. (3.60):

a2 1 3 a4

T1 cos 2
2
2 2 r4
r
rr(2) = 2
1
2
(2) = +
3 a4
T1 cos 2
2 r4
(3.65)
a2 1 3 a4
+
sin 2
2
2 2 r4
r
r(2)
=
In order to obtain the total solution, we reevaluate Eq. (3.57), by dividing the numerator
and denominator of each term by b2, letting a / b 0 , and then adding the results to Eq.
(3.65):
T1 a 2 T1
a2
a4
rr ( r , ) = 1 2 + 1 4 2 + 3 4 cos 2
r
r
2 r 2
T a2 T
a4
( r , ) = 1 1 + 2 1 1 + 3 4 cos 2
2 r 2
r
r ( r , ) =
(3.66)
T1
a2
a4
1
2
3
+
sin 2
2
r2
r4
Now we may examine the stresses at particular points that are most important. For
instance, the maximum hoop stress that occurs at r = a is:
( a, ) = T1 2T1 cos 2
Note that its largest value occurs at =

a, = 3T1
2
(3.67)
,
with the magnitude of:
2 2
(3.68)
which indicates that we have a stress concentration factor of 3 for the specified location.
Assignment 9:
Find the stress concentration factor for a small circular hole in a large sheet subjected to a
remote tensile stress T1 and a compressive stress T2 .
- 81 -
General Solution of Plane Problems using Complex

Variable Functions
In this section we will use complex variable functions to obtain general solutions to the
plane elasticity problems. Starting with the biharmonic equation we can show:
4 = 0
4 = 2 2 = (
),
4
4
4
+
+
2
x14
x12 x22 x24
2
2 2 2
2 + 2 2 + 2 = 0
x1 x2 x1 x2
2
2
2 + 2 ( 22 + 11 ) = 0
x1 x2
(3.69)
2P 2P
2 + 2 = 2 P = P, = 0
x1 x2
The conclusion is that that P is harmonic. Now we assume that there is a complex
variable function which has P as its real part:
f ( z ) = P + iQ
(3.70)
Since f ( z ) is analytic, its integral is also be analytic. Accordingly, we may write:
f ( z )dz = ( P + iQ ) dz =
lets say 4( z )
( z ) = p + iq
1
f ( z )dz
4
d ( z )
,1 =
z,1 = ( z )
dz
1
( z ) = p,1 + iq,1 = f ( z )
4
1
P
Q
p,1 + iq,1 = ( P + iQ ) p,1 =
, q,1 =
4
4
4
( z ) =
Invoking the Cauchy-Riemann conditions we will have:
(3.71)
- 82 -
p,1 = q,2 = 4 (i)
p = q = Q
,1
,2
4
2
(i) = P = 4 p,1 = 2 p,1 + 2q,2
(3.72)
2 2 p,1 2q,2 = 0
2 ( x1 p x2 q ) = 0
Note that ( x1 p x2 q ) is a solution of the Laplaces equation, so we may assume it to

be the real part of a complex variable function:
( x1 p x2 q ) = p1
= x1 p + x2 q + p1
(3.73)
where we have:
( z ) = p1 + iq1
(3.74)
Moreover, = x1 p + x2 q + p1 may be assumed to be the real part of the following analytic

function:
( x1 ix2 )( p + iq ) + p1 + iq1 or
= x1 p + x2 q + p1 = e ( x1 ix2 )( p + iq ) + p1 + iq1
(3.75)
Finally, we may write the general form of the Muskhelishvili stress function in terms of
two complex potentials and as follows:
= e [ z ( z ) + ( z ) ]
(3.76)

Starting with the constitutive equations for plane stress we have:
1 +
= +

E
which may be rewritten as:
(3.77)
- 83 -
Eu1,1 = 11 22
(i)
Eu2,2 = 22 11
(ii)
G ( u1,2 + u2,1 ) = 12
(iii)
(3.78)
Now, using Eqs. (3.72) we may write:

Eu1,1 = (1 + ),11 + P
Eu2,2 = (1 + ),22 + P
P = 4 p,1 = 4q,2
4
2Gu1,1 = ,11 + 1 + p,1
2Gu = + 4 q
2,2
,22
,2
1 +
(3.79)
The above equations can be integrated to give the displacements:
4
2Gu1 = ,1 + 1 +
2Gu = + 4
2
,2
1 +
p + f ( x2 )
q + f ( x1 )
(3.80)
If we substitute the above equations into Eq. (3.78iii), it can be shown that f(x2) and f(x1)
are constants. Now we may substitute for the with our general complex variable stress
function (Eq. (3.76)) but first we note:
f ( z ) = + i
f ( z ) + f ( z ) = 2 = 2e ( f ( z ) )
f ( z ) = i
(3.81)
so from (3.76) z ( z ) + ( z ) + z ( z ) + ( z ) = 2
If we substitute the above expression for the into Eq. (3.80), we will obtain the
following general expression for the displacements in terms of the two complex
potentials and :
2G (u1 + iu2 ) =
3
( z ) z ( z ) ( z )
1 +
(3.82)
It should be noted that we can obtain the above equation for the plane strain condition by
replacing with
- 84 -
Having found the displacements, the general expressions for the stresses can be easily
obtained as follows:
11 + 22 = 4e ( ( z ) )
22 11 2i 12 = 2 z ( z ) + ( z )
22 11 + 2i 12 = 2 [ z ( z ) + ( z )]
or
(3.83)
The final conclusion is that the solution to a particular two-dimensional elasticity

problem can be easily obtained if we can find the appropriate complex variable
potentials for that problem. However, finding the required potentials is not usually
that easy!
The Mode I Crack Problem

As an example for the solution of plane problems using complex variable potentials, we
consider the Mode I crack problem. Here, we may assume the displacement field as u1 =
u1(x1, x2), u2 = u2(x1, x2), and u3 = 0.
Fig. 3.6
We directly start from the general solution with the following form of Eq. (3.83):
11 + 22 = 4e ( ( z ) ) = 2 ( z ) + ( z )
22 11 2i 12 = 2 z ( z ) + ( z )
(3.84)
Due to symmetry with respect to the crack plane we choose a solution of the form:
= Az +1 , = Bz +1
(3.85)
- 85 -
where A, B, and are real constants. In order to have nonsingular displacements at the
crack tip we must have > 1. The substitution of Eq. (3.85) into Eq. (3.84) yields:
22 i 12 = ( + 1)r { A [ 2 cos + cos( 2) ]

+ B cos i [ A sin( 2) + B sin ]}
(3.86)
The boundary conditions at the crack tip dictate that the stresses in Eq. (3.86) vanish
for = . Consequently we have:
A(2 + ) cos + B cos = 0
A sin + B sin = 0
(3.87)
for which a nontrivial solution exists if:

sin 2 = 0
(3.88)
which for > 1 has the following roots:

1
2
= , n / 2, n = 0,1, 2,...
(3.89)
The dominant contribution to the crack-tip stress and displacement fields occurs for
= 1/2 for which A = 2B. An inverse square root singularity in the stress field exists at
the crack tip. Substituting Eq. (3.87) with A = 2B and = 1/2 into Eqs. (3.86) and
(3.82), we can show that:
11
1 sin( / 2) sin(3 / 2)
KI

cos( / 2) sin( / 2) cos( / 2)

12 =
1/ 2
(2 r )
1 + sin( / 2) sin(3 / 2)
22
(3.90)
and also:
1
2
u1 K I r 2 cos( / 2) 1 + 2sin ( / 2)
=

2
u2 2 2 sin( / 2) + 1 2 cos ( / 2)
(3.91)
The Mode I stress intensity factor K, which can be obtained using the global boundary
conditions of the problem, is defined by:
K I = lim {(2 r )1/ 2 22
r 0
=0
(3.92)
- 86 -
General Solutions in Curvilinear coordinates

For many two-dimensional elasticity problems the solution procedure requires that we
define our previously derived general solutions using curvilinear coordinates. In general,
we may specify the position of a point in a plane as the intersection of two curves. This
constitutes the curvilinear coordinates. The two curves may be represented by:
F1 ( x1 , x2 ) =
F2 ( x1 , x2 ) =
(3.93)
X2
=constant
=constant
X1
Fig. 3.7
For instance, the usual polar coordinates can be considered as a form of curvilinear
coordinates which specify the position of our point as an intersection of a radial line, at
the angle from the initial line, with a circle of radius r. Then, we have:
F ( x , x ) = x2 + x2 = r
1
2
1 1 2
x2
F2 ( x1 , x2 ) = arctan =
x1
(3.94)
We may also solve Eqs. (3.93) for x1 , x2 and obtain:

x1 = f1 ( , )
x2 = f 2 ( , )
(3.95)
By choosing different functions for f1 and f2 we can find various curvilinear coordinates.
Let us assume:
x1 = f1 ( , ) = C cosh cos
x2 = f 2 ( , ) = C sinh sin
(3.96)
- 87 -
If we eliminate between the above two equations, we can show that:

x12
x22
+
=1
C 2 cosh 2 C 2 sinh 2
(3.97)
which for different values of gives different ellipses with the same foci, i.e., a family
of confocal ellipses. Alternatively, if we eliminate between the two equations, we will
have:
x12
x22
=1
C 2 cos 2 C 2 sin 2
(3.98)
which for different values of gives different hyprbolas with the same foci, or a family
of confocal hyperbolas.
X2
X1
Fig. 3.8
We may also define the complex variable = + i in terms of our curvilinear
coordinates by writing the following equalities using Eq. (3.96):
x1 + ix2 = C cosh( + i )
z = C cosh
(3.99)
z = f ( )
Now we express the stresses in the curvilinear coordinates in terms of the stresses in the
cartesian coordinates by a simple transformation caused by a rotation of degrees:
11 + 22
- 88 -
11 22
cos 2 + 12 sin 2
2
2
+ 22 11 22
cos 2 12 sin 2
= 11
2
2

= 12 cos 2 11 22 sin 2
2
+
(3.100)
which leads to:
+ = 11 + 22
+ 2i = e 2i ( 22 11 + 2i 12 )
(3.101)
Finally, we use Eqs. (3.83) and (3.82) and write the general expressions for the stresses
and displacements in terms of complex potentials in the curvilinear coordinates as:
+ = 4e ( ( z ) )
+ 2i = 2e 2i [ z ( z ) + ( z )]
(3.102)
( z ) z ( z ) ( z )
2G (u iu ) = ei
1 +
In order to find the factor e 2i , we write:

f ( ) =
df ( )
d
f ( ) = j cos + ij sin = jei

f ( ) = je
f ( )
e2i =
f ( )
(3.103)
For example, for the elliptic coordinates we can show that:

e2i =
sinh
sinh
(3.104)
- 89 -
Uniformly Stressed Sheet with an Elliptic Hole

Consider a large sheet, containing an elliptic hole with semiaxes, a and b, subjected to a
far-field uniform stress 0 as depicted in Fig. 3.9. We may define our elliptic coordinates
as:
z = C cosh
; = + i ;
x1 = C cosh cos
x2 = C sinh sin
dz
= C sinh
d
(3.105)
C cosh 0 = a
C sinh 0 = b
0
X2
0
X1
2C
Fig. 3.9
For this case, it is appropriate to use the general expressions for the stresses and
displacements in terms of complex potentials in curvilinear coordinates (Eqs. (3.102))
with the following boundary conditions:
x1 , x2 ( ) 11 = 22 = 0 (i)
= 0 = = 0
From Eqs. (3.102) we have:
(ii)
(3.106)
- 90 -
+ = 4e ( ( z ) )
(i)
+ 2i = 2e 2i [ z ( z ) + ( z )]
(ii)
(3.107)
which, in combination with our first boundary condition, indicates that we must have:
+ = 4e ( ( z ) ) = 2 0 e ( ( z ) ) = 0
2
z ( z ) + ( z ) = 0
at infinity
(3.108)
In order to find the appropriate complex potentials, we note that they should conform to
our elliptic boundary and be periodic in . Hence, we assume the following potentials
and check if our boundary conditions are satisfied:
( z ) = AC sinh
2
( z ) = BC
( z ) = AC cosh
d
cosh
=A
= A coth
dz
sinh
coth 1 e( z ) = A =
BC
cosh
( z ) = B
sinh
sinh 3
A 1
( z ) =
C sinh 3
z ( z ) + ( z ) = 0
0
2
(3.109)
( z ) =
It is clear that with A =
the far field boundary conditions are satisfied. In order to

2
find B, we proceed with the local boundary conditions. From Eq. (3.107) we have:
(i) (ii) i = 2e ( ( z ) ) e2i [ z ( z ) + ( z ) ]
i = ( z ) + ( z ) e 2i [ z ( z ) + ( z )]
(3.110)
where e 2i can be found from Eq. (3.104) as:

e2i =
sinh
sinh
Accordingly, Eq. (3.110) becomes:
(3.111)
- 91 -
cosh cosh
+
i = A
sinh sinh
sinh cosh
cosh
+B
+
A
3
sinh 3
sinh sinh
1
A sinh sinh ( + ) + cosh + B cosh
=
2
sinh sinh
(3.112)
which may be written at the boundary of the ellipse as:
= 0
+ = 2 0 = 2 0
1
=
( A cosh 20 + B ) cosh
2
sinh sinh
(3.113)
Invoking the second boundary condition in Eq. (3.106ii) we have:

1
B = A cosh 2 0 = 0 cosh 2 0
2
1
( z ) = 0C sinh
( z ) = 1 0 cosh 2 0 C 2
2
(3.114)
Although all the boundary conditions are now satisfied, we must make sure that there
would be no discontinuity in the resulting displacement field. Hence, we use Eq. (3.102)
to obtain the Cartesian components of displacement as:
2G (u + iv) =
3
BC
AC sinh AC cosh coth
1 +
sinh
(3.115)
which shows that displacements are periodic in and there would be no discontinuity in
displacements. The most interesting component of the stress is in fact the largest value of
at the hole and can be obtained from Eq. (3.107) as follows:
+ = 4e ( ( z ) ) = 2 0e coth
2 0 sinh 2
sinh 2 i sinh 2
+ =
cosh 2 cos 2
cosh 2 cos 2
2 0 sinh 2 0
( ) = =
0
cosh 2 0 cos 2
coth =
( ) =0, =
max
2 0 sinh 2 0
cosh 2 0 1
(3.116)
- 92 -
Using Eq. (3.105) we may show that:

C 2 = a 2 b2
2ab
sinh 2 0 = 2
C
a 2 + b2
cosh 2 0 =
C2
(3.117)
which, in combination with Eq. (3.116), results in:
( )
max
( )
min
= 0,
= 2 0
= ,
2 2
a
b
= 2 0
b
a
(3.118)
Eqs. (3.118) show that as b 0 (as the ellipse becomes a crack) a stress singularity
develops at the crack tip. For a=b, the ellipse becomes a circle with a stress concentration
factor of two, which is in agreement with the results we already obtained for the sheet
with a circular hole.
Assignment 10:
Find the maximum stresses for an elliptic hole in a large sheet subjected to remote tensile
stress T1 . The major axis of the elliptic hole makes an angle with the x1 direction.
- 93 -
Chapter 4
Linear Thermoelasticity
Temperature change can often be a source of significant stresses in elastic bodies. In
general, we should define a reference temperature at which we consider the body to be
free of thermal stresses. As the temperature is raised or lowered with respect to the
reference temperature the body expands or contracts and thermal stresses may arise.
It should be note that for a body which is not constrained, a uniform temperature change
with respect to the reference temperature will not cause thermal stresses. However, this
is not the case for a constrained body, where the thermal stresses are developed even if
the temperature changes are uniform throughout the body. In either case, the existence of
temperature gradients within an elastic body will result in the development of thermal
stresses which are generally tensile at locations of lower temperatures and compressive at
locations of higher temperatures. Although later we will examine some special cases
where the thermal stresses can be absent also for bodies under temperature gradients.
Usually, the solution of the thermoelasticity problem begins with finding the time and
position dependent temperature field. We may write the general form of the heat
conduction equation as:
S
= a,ii +
Cv
t
; a=
k
Cv
(4.1)
in which is the temperature field , S is the strength of the heat sources in the field ,
is the density, Cv is the specific heat, and k is the thermal conductivity. For a stationary
field with no heat sources we have:
,ii = 0
(4.2)
Since we are dealing with linear thermoelasticity, the strain tensor is defined by:
ij =
1
( ui, j + u j ,i )
2
(4.3)
and also the equilibrium equations are:
ij , j + fi = 0
(4.4)
However, we may split the total strain tensor into the mechanical and thermal parts as:
- 94 -
ij = ij m + ij T
(4.5)
For the mechanical stresses we have:

1 +
ij kk ij
E
E
E
G=
2 (1 + )
ij m =
ij m =
(4.6)
kk ij
ij
2G
1 +
In order to obtain an expression for the thermal strains we may write:

dST = (1 + ) dS0
(4.7)
Here dS0 is the initial length of a line element, dST is its length as a result of a
temperature change , and is the coefficient of thermal expansion. Accordingly, the
thermal strains are:
ij T = ij
(4.8)
We can combine Eqs. (4.6) and (4.8) to obtain the total strains, resulting in the so-called
Duhamel-Neumann Constitutive equations:
ij = ij m + ij T =
kk ij + ij
ij
2G
1 +
(4.9)
In order to find the stresses in terms of the strains, we first contract on i and j indices to
obtain:
1 1 2
ll + 3
2G 1 +
2G (1 + )
ll =
( ll 3 )
1 2
ll =
Substituting back into Eq. (4.9) we have:
(4.10)
ij =
- 95 -
1
2G
ij
( ll 3 ) ij + ij
2G
1 2
1
+1
=
ij
ll ij +
ij
2G
1 2
1 2
(4.11)
Hence, the stresses are:
ij = 2G ij +
1 2
2G (1 + )
ij
1 2
ll ij
(4.12)
Based on our new set of constitutive equations and implementing the procedure we
followed before, we may obtain a generalized form for the Navier displacement
equations as follows:
ui , jj +
2 (1 + )
1
1
u j ,ij + f i =
G
1 2
(1 2 ) ,i
(4.13)
We may also derive a generalized form for the Beltrami-Michell equations:
(1 + ) ij ,kk + kk ,ij = (1 + ) fi , j + f j ,i +
f k , k ij
1
1 +
2G (1 + ) ,ij +
, kk ij
1
(4.14)
A Thin Circular Disc Problem

As a first example in solving thermoelasticity problems, we will find the thermal stresses
in a thin circular disc. We assume that the temperature field is symmetrical about the
center of the disc and does not vary over the thickness. For this case the equilibrium
equations simplify to:
d rr rr
+
=0
dr
r
(4.15)
We now repeat the procedure implemented above for determination of the thermoelastic
constitutive equations for our specific problem. Starting with the strains we write:
1
( rr )
E
1
= ( rr )
E
which can be written in terms of the stresses as:
rr =
(4.16)
- 96 -
E
rr + (1 + )
1 2
E
=
+ rr (1 + )
1 2
rr =

(4.17)
Substituting the above expressions into our equilibrium equation (Eq. (4.15)) we have:
r
d
d
( rr + ) + (1 )( rr ) = (1 + ) r
dr
dr
(4.18)
which can be rewritten in terms of displacements to give:

d 2ur 1 dur ur
d
+
2 = (1 + )
2
dr
r dr r
dr
d 1 d (rur )
d

= (1 + )
dr r dr
dr
(4.19)
This is in fact nothing but the simplified form of the generalized Navier equations for the
current problem. The above differential equation can be solved for the radial
displacement to give:
ur = (1 + )
C
1 r
rdr + C1r + 2
r a
r
(4.20)
Accordingly, the stresses become:

1
C1 (1 + ) C2 (1 ) r 2
1 r
E
1
= E 2 rdr E +
C 1 + ) C2 (1 ) 2
2 1(
a
r
1
r
rr = E

1
r2
rdr +
E
1 2
(4.21)
The constants can be found from the boundary conditions. For a solid disc, a can be
taken as zero and we also have:
1 r
rdr = 0
r 0 r 0
lim
(4.22)
Note that Eq. (4.20) implies that we must have C2 = 0 to ensure that the displacements at
the center of the disc have finite values. Moreover, at the outer edge of the disc we
have rr = 0 , so we may write:
C1 = (1 )
- 97 (4.23)
rdr
Finally, the stresses can be shown to be:

1
2
b
rdr
1 b
1 r
= E + 2 rdr + 2 rdr
0
0
b
r
rr = E

rdr
1
r2
(4.24)
Assignment 11:
Find the thermal stresses for a hollow sphere for which the inner and outer surfaces are
kept at constant temperatures Ti and To respectively.
Instantaneous Point Source in an Infinite Body

In this example we consider a quantity of heat Q deposited at time t = 0 at Xi = 0 inside an
infinite linear elastic body. The heat will gradually spread into the body, and every point
will experience an increase and later decrease in temperature with time. Eventually, the
body will return to its initial temperature after a long period of time. In order to obtain
the resulting thermal stresses, we must first find the time and position dependent
temperature field. The problem can be treated as spherically symmetric, so we may start
with the proper form of the equation of heat conduction in spherical coordinates:
1
= a 2 r2
t
r r r
(4.25)
Defining the Laplace transform as:

( s ) = (t )e st dt
(4.26)
we have:
= s
t
The Laplace transform of Eq. (4.25) results in:
(4.27)
=
s
- 98 (4.28)
a d 2 d
r
2
r dr dr
The solution to the above differential equation can be shown to be:

s
r
a
= A e
s
r
a
B
+ e
r
(4.29)
remains bounded as r .
Note that we should set B = 0 to ensure that
In order to find the constant A, we write:
(4.30)
Q = 4 r 2 Cv dr
0
We also define Q using the Heaviside step function as:
Q = QH (t )
(4.31)
Next, we take the Laplace transform of both sides of Eq. (4.30) and proceed as follows:
1
dr
Q = 4 r 2 Cv
s 0
= 4 Cv A e
s
r

a
= 4 Cv A
rdr
(4.32)
a
s
Accordingly, we find the constant A as:

A=
(4.33)
Q
4 Cv a
Finally, we find the time and position dependent temperature field for our problem as
follows:
=
s
r
a

Q
e
4 Cv ar
Q
3/ 2
( 4 at ) Cv
(4.34)
e
r2
4 at
- 99 -
The next step in solving our thermoelasticity problem is the determination of the
displacement field. We start with the generalized form of the Navier equations (Eqs.
(4.13)):
ui , jj +
2 (1 + )
1
1
u j ,ij + f i =
G
1 2
(1 2 ) ,i
(4.35)
The solution to the above equation consists of the general solution uih plus the particular
integral uip . Now, we may assume that the particular solution can be derived from the
thermoelastic potential as follows:
(4.36)
uip = ,i
If we substitute the above expression into the Generalized Navier equations, we will
have:
2 (1 + )
1
, jj
= 0
, jj +
1 2
1 2
,i
(1 + )
, jj = 2 =
(1 )
(4.37)
Assuming a non-stationary source-free temperature distribution, a solution for the above

equation can be written as:
=
(1 + ) a t dt + + t
0
1
(1 ) 0
(4.38)
where 2 1 = 0 , and 0 represents the initial thermoelastic potential.

We now turn our attention to the homogeneous form of the Navier equations. For the
current problem, since there are no body forces or surface tractions, we may assume:
(4.39)
uih = 0
Substituting the obtained temperature field into the general solution for the thermoelastic
potential, we can write:
=
r2
4 at
t e
(1 + ) Q a
3/ 2
(1 )( 4 a ) Cv 0 t 3/ 2
dt + 0 + t 1
In order to ease the formulations, we may further consider;
(4.40)
- 100 -
r
r2
2
=
=
V
V
4at
2 at
r
r3
3/ 2
=
=
t
t
2
8a 3/ 2V 3
4aV
r2
dt =
dV
2aV 3
(4.41)
which upon substitution into Eq. (4.40), results in:

1 + ) Q 2 2 rat V
(
0 t1 =
e dV
(1 ) 4 r Cv
(1 + ) Q 2 0 eV dV + 2 eV dV
=
r
(1 ) 4 r Cv 2 at
0
r
(1 + ) Q 2 2 at eV dV + 2 1
=
(1 ) 4 r Cv 0
2
(1 + ) Q 1 erf r
=
(1 ) 4 r Cv
2 at
2
(4.42)
The error function is:

r
erf
2 at
r
2 at
0
e V dV
(4.43)
Now we will determine 1 and 0 values of the thermoelastic potential. After a long
period of time the displacements and stresses return to zero, so we may write:
1 = 0
(1 + ) Q
=
0
t =
(1 ) 4 r Cv
(1 + ) Q erf r
=
(1 ) 4 r Cv 2 at
(4.44)
Having found the thermoelastic potential, the displacement, strain, and stress fields can
be determined as follows:
- 101 -
ui = uih + uip = ,i
ij = ,ij
1 2
ij = 2G ,ij +
,kk ij
(1 + )
1 2
(4.45)
ij
Two-Dimensional Thermoelastic Problems

Plane strain
Here we start by reviewing some of the plane strain formulations:
u1 = u1 ( x1 , x2 )
u = u ( x )
u2 = u2 ( x1 , x2 )
u3 = 0, constant
11
1
ij = 2 ( ui , j + u j ,i )
, + f = 0
f = V,
11 V = ,22
22 V = ,11
12 = 21 = ,12
, = 1, 2
12 0
22 0
(4.46)
The two-dimensional form of the Duhamel-Neumann constitutive relations is:
1 2
= 2G +
(1 + )

1 2
3 = 2G 3 = 0

(1 + )

1 2
1 2
33 = 2G
(i)
(ii)
(4.47)
(iii)
In order to find the appropriate expression for 33 we need to find . We start by

contracting on and indices in Eq. (4.47i), as follows:
= 2G +

- 102 -
2
2 (1 + )

1 2
1 2
1 2
=
+ 2 (1 + )
2G
(4.48)
Substituting for in Eq. (4.47iii) results in:
(1 + )
2G
33 = E
33 = 2G
(4.49)
which indicates that 33 may be determined from a knowledge of and .

Furthermore, we have:
1
+ (1 + )
2G
(4.50)
which is the two-dimensional form of the Duhamel-Neumann relation. Now, using the
above expression, and implementing the same approach we followed in Chapter 3, we
can derive the biharmonic equation for linear thermoelasticity as follows:
1 2 2
1 + 2
4 =
V 2G

1
1
(4.51)
If body forces are negligible or constant we may write:

1 +
4 = 2G
1
(4.52)
The solution to the above biharmonic equation is:
= h + p
(4.53)
where h can be found from the elasticity part of the problem without considering the
thermal effects. In order to find the particular integral, we use the definition of the
thermoelastic potential, Eqs. (4.36), in two-dimensional form:
up = ,
which satisfies the two-dimensional form of Eq. (4.37):
(4.54)
, = 2 =
- 103 -
(1 + )
(1 )
(4.55)
If we differentiate the above equation twice we will have:
4 =
(1 + ) 2
(1 )
(4.56)
Substituting into our biharmonic equation (Eq. (4.52)) we can write:

4 p = 2G4
p = 2G
(4.57)
Naturally, the isothermal deformation satisfies the homogeneous form of Eq. (4.52), and
as a result the complete solution for the stress function is:
= h + p = h 2G
(4.58)

We start by reviewing some of the plane stress formulations:
ij =
ij
1 h/2
ij dx3
h h / 2
11 12 0
22 0

0
(4.59)
The bar sign over a quantity indicates that it has been averaged over the thickness. The
equilibrium equations can be written as:
f = V,
, V, = 0
(4.60)
which are exactly the same as those derived for plane strain condition, except for the
mean value interpretation, so we may similarly write:
- 104 -
11 V = ,22
22 V = ,11
(4.61)
12 = 21 = ,12
The constitutive equations are:
1
2G
1 +
1
3 =
3 = 0
2G
1
33 =
+
2G 1 +
(4.62)
which can be inverted in a standard manner to give:
2G
(1 ) + (1 + )
1
(4.63)
Now, similar to our plane strain formulation, upon substitution for the strains into the
only nonzero compatibility equation we find that:
(4.64)
4 = E2
The solution to the above biharmonic equation is:
= h + p
(4.65)
where h can be found from the corresponding isothermal elasticity problem. In order to
find the particular integral, we use the plane stress form of the thermoelastic potential
expression to write:
2 = , = (1 + )
4 = (1 + ) 2
using (4.49) 4 =
p = 2G
E
4
(1 + )
(4.66)
- 105 -
Solution Strategies for Two-Dimensional Problems

As mentioned before, the solution of the thermoelasticity problems begins with finding
the time and position dependent temperature field. For the two-dimensional problems,
the next step is the solution of either of the following expressions for the thermoelastic
potential:
2 = , =
(1 + )
(1 )
; 2 = , = (1 + )
(4.67)
which using p = 2G will give the particular solution.

The solution to the isothermal elasticity problem can be found using the methods
introduced earlier, or it can be constructed as a finite or infinite series of properly chosen
biharmonic functions. In Cartesian coordinates we may use:
e x2 cos x1 ; x1e x2 cos x1 ; x2 e x2 cos x1
(4.68)
where is an arbitrary constant. All other combinations obtained by interchanging x1 and

x2 and/or replacing cos x1 by sin x1 can also be used.
In Polar coordinates, a complete set of biharmonic functions which are periodic in are:
r 2 ; log r ; r 2 log r
(4.69)
r log r cos ; r n cos n ; r 2 n cos n ; n = 0,1, 2,K
and the corresponding expressions with cos n replaced by sin n can also be used.
Assignment 12:
Using the stress function approach, find the thermal stresses for a long circular hollow
cylinder for which the inner and outer surfaces are kept at constant temperatures Ti and
To respectively. The cylinder is also under internal pressure Pi.
Two-Dimensional Simply-Connected Regions with Steady Heat Flow

Here, we will show that there are special cases where the thermal stresses can be absent
even if the body experiences temperature gradients. Let us consider a steady heat flow
which is confined to the x1x2 plane, for example, consider a long cylinder with no
variation of temperature in the axial direction. The temperature field will satisfy the 2D
form of the Laplaces equation:
- 106 -
, = 0
(4.70)
and the constitutive equations are:
11 (1 + ) =
1 2
22
11
E
1
1 2
22 (1 + ) =
11
22
E
1
(4.71)
Now, in order to maintain 11 = 22 = 0 , we should have:
11 = (1 + )
(4.72)
22 = (1 + )
The above strains should satisfy the compatibility equations, which in this case simplify
to:
11,22 + 22,11 = 0
(4.73)
which is identically satisfied if we substitute Eqs. (4.72) into Eq. (4.70). Hence we may
conclude that the thermal stresses can be absent for two-dimensional simply-connected
regions with steady heat flow (even for bodies under temperature gradients). As we will
see in the next example the situation is different for non-simply-connected regions.
Insulated Circular Hole inside an Infinite Plate subjected to Uniform

Heat Flow
In this example we assume a uniform heat flow in the direction of the negative x2 axis in
an infinite plate. For this case we may write:
= qx2
(4.74)
where q is the constant temperature gradient. As depicted in Fig. 4.1, the uniform heat
flow is disturbed by the presence of an insulated circular hole.
- 107 -
X2
X1
Fig. 4.1
In order to obtain the thermal stresses, we should first find the temperature field. The
two-dimensional form of Eq. (4.2) for plane stress is:
(4.75)
, = 0
which can be written in cylindrical coordinates as:

2 1 1 2
+
+
=0
r 2 r r r 2 2
(4.76)
We may solve the above partial differential equation using the separation of variables
method as follows:
= R(r ) X ( )
(4.77)
Accordingly we may write:

dR
X
=
r dr
2 d 2 R
= 2 X
dr
r 2
dX
=R
d
2
d2X
;
R
=
d 2
2
;
Substituting the above expressions into Eq. (4.76), we have:
(4.78)
- 108 -
d 2R
1 dR
R d2X
=0
X+
X+ 2
dr 2
r dr
r d 2
r 2 d 2 R r dR 1 d 2 X
+
+
=0
R dr 2 R dr X d 2
(4.79)
which upon splitting into two parts results in:

r2
R
1
X
d 2 R r dR
+
= k2
2
dr
R dr
2
d X
= k 2
d 2
(i)
(4.80)
(ii)
Equation (4.80i) can be rearranged as the Cauchy-Eulers equation:

r2
d 2R
dR
+r
k 2R = 0
2
dr
dr
(4.81)
which can be solved using the following considerations:

r = e z ; z = ln r
dR dR dz 1 dR
=
=
dr dz dr r dz
d 2R
1 dR 1 d 2 R
2 = 2
+
dr
r dz r 2 dz 2
(4.82)
With proper substitutions into Eq. (4.81), the solution can be obtained as follows:
d 2R
= k 2R
2
dz
R = A1e kz + A2 e kz
(4.83)
R = A1r k + A2 r k
Now we turn our attention to Eq. (4.80ii), which can be solved as the Eulers equation:
d2X
+ k2X = 0
d 2
X = B1eik + B2 eik
X = C1 sin k + C2 cos k
Finally, the temperature field can be written as:
(4.84)
- 109 -
= RX = ( A1r k + A2 r k ) ( C1 sin k + C2 cos k )
(4.85)
for which the four constants can be determined from the boundary conditions. Also note
that we have:
= k ( A1r k 1 A2 r k 1 ) ( C1 sin k + C2 cos k )

r
(4.86)
The first temperature boundary condition is:
= 0 ; = 0 r
(4.87)
C2 = 0
Thus, Eqs. (4.85) and (4.86) can be rewritten as:
= ( A1C1r k + A2C1r k ) sin k = ( D1r k + D2 r k ) sin k
= ( D1r k 1 D2 r k 1 ) k sin k
r
(4.88)
The second temperature boundary condition is:
= 0
r
D2 = D1a k 1+ k +1 = D1a 2 k
r=a ;
(4.89)
This time, Eq. (4.88) can be rewritten as:

a 2k
= D1r 1 + 2 k sin k
r
a 2k
= D1kr k 1 1 2 k sin k
r
r
k
(4.90)
The third temperature boundary condition is:

=
=q
r x2
a 2k
D1kr k 1 1 2 k
r
=q
The above expression must remain valid as r so we may conclude that:
(4.91)
- 110 -
k 1 = 0 k = 1
D1 = q
(4.92)
Finally, the temperature distribution can be found as:
a2
= q r sin + sin
r
a2
= q x2 + sin
r
(4.93)
The above expression shows that the uniform temperature field qx2 remains unchanged at
a2
sin term. As
r
discussed before, the uniform temperature field has no influence on the thermal stresses
and may be disregarded in the process of determination of the thermoelastic potential
from its governing equation as follows:
the far field and the effect of the circular hole is represented by the q
2 = (1 + )
a2
2 1 1 2
q
1
sin
+
+
=
+
(
)
r
r 2 r r r 2 2
1 + ) qa 2
(
r log r sin
=
2
(4.94)
Having found the thermoelastic potential, we may easily calculate p from p = 2G .

In order to have some clue to the proper form of h , we may proceed as follows:
= h + p = h 2G
rr = rrh + rrp
1 1 2
= rrh 2G
+ 2
2
r r r
2G (1 + ) qa 2 1
h
= rr
sin
2
r
E qa 2
=
sin
2r
h
rr
Now we may apply our first overall stress boundary condition:
(4.95)
- 111 -
rr = 0 at r = a
E qa
rrh
=
sin
r =a
2
1 1 2 h
+ 2
2
r r r
r =a
E qa
=
sin
2
(4.96)
which suggests the following form for h :
h =
E qa
f (r ) sin ; f (a) = 1
2
(4.97)
Similarly, for the shear stresses we have:
r = rh + rp
1
= rh 2G
r r
E qa 2
= rh +
cos
2r
(4.98)
Applying our second overall stress boundary condition, we will have:
r = 0 at r = a
E qa
rh
=
cos
r =a
2
1 h

r r
r =a
E qa
=
cos
2
(4.99)
which approves the previously suggested form for h . However, according to our third
overall boundary condition all the stresses should vanish as r . Accordingly, we
may modify the expression (4.97) as:
h =
E qa 1
f sin ; f (a ) = 1
2
r
(4.100)
Here, we consult the list of the biharmonic functions in polar coordinates and finalize our
expression as:
h =
E qa 1
Ar sin
2
Accordingly, we can write:
(4.101)
E qa
sin
r =a
2
1 1 2 h
AE qa
sin
r =a =
+ 2
2
a3
r r r
rrh
- 112 -
(i)
a3
(i),(ii) A =
2
(ii)
E qa a
=
sin
2 r
(4.102)
h
rr
rr =
E qa a a 3
sin
2 r r3
The other stress components can be calculated using h as follows:
h =
r
E qa 4
sin
4r
E qa a a 3
=
cos
2 r r3
(4.103)
E qa a a 3
+ sin
2 r r3
Zero Thermal Stresses in a Three-Dimensional Body

As mentioned before, there are some special cases where the thermal stresses can be
absent, even if the body experiences temperature gradients. Previously, we examined the
steady heat flow in two-dimensional simply-connected regions. We may also arrive at
zero thermal stresses for an unrestrained body when the temperature distribution is a
linear function of rectangular space coordinates. For the stresses all to be zero, the six
stress compatibility equations result in:
,11 = ,22 = ,33 = ,12 = ,23 = ,31 = 0
(4.104)
Hence, the thermal strain distribution must be a linear function of the rectangular space
coordinates like:
= a0 + a1 x1 + a2 x2 + a3 x3
(4.105)
- 113 -
Note that the coefficients may be time dependent. Accordingly, if the thermal properties
of the material are constant, the zero stress condition dictates a linear temperature
distribution.
Assignment 13:
Consider an unrestrained circular disc for which we have = Ar . Based on our previous
discussions concerning the special cases for zero thermal stress, can we conclude that the
disc is stress free? Why?
- 114 -
Chapter 5
Variational Principles and Applications
Leonhard Euler: Since the fabric of the universe is most perfect, and is the work of a most
wise creator, nothing whatsoever takes place in the universe in which some relation of
maximum and minimum does not appear. Wherefore, there is absolutely no doubt that every
effect in universe can be explained as satisfactorily from final causes, by the aid of the method
of Maxima and Minima, as it can from the effective causes themselves. Ref: Timoshenko, S.,
History of Strength of Materials (McGraw-Hill Book Company, Inc., New York, 1953)
Energy Principles
As clearly stated by Leonhard Euler in the 18th century, the field equations of the theory
of elasticity may also be developed from the energy considerations. In particular, these
methods provide powerful tools for obtaining approximate solutions and implementation
of numerical solution techniques.
Principle of Virtual Work

The virtual work can be defined as the work done on a deformable body, by all the
forces acting on it, as the body is given a small hypothetical displacement which is
consistent with the constraints present. The virtual displacements are represented by the
symbol , known as the variation operator. In general, the loadings consist of body
forces and surface tractions. The later are prescribed over a part of the boundary
designated by S . Over the remaining boundary (designated by Su ) the displacement
field u is prescribed. However, it must be ensured that u = 0 on Su to avoid violating
the constraints.
T
Su
Su
f
Fig. 5.1
Accordingly, the virtual work is defined by:
- 115 -
Wvirt = T. u + f . u
S
(5.1)
= Ti ui dS + f i ui dV
S
Now we may invoke the stress boundary relations and implement the divergence theorem
to obtain:
T u dS + f u dV =
i
ij
n j ui dS + fi ui dV
V
= ( ij ui ) + f i ui dV
,j
(5.2)
=0
= ( ij , j + f i ) ui + ij ( ui , j ) dV
The first grouping of terms within the last integral can be set to zero to ensure the
equilibrium. The product of the symmetric stress tensor with the skew-symmetric part of
ui , j is also zero. Since the symmetric part of ui , j is nothing but a variation in the
strain tensor, we may write the following expression, known as the principle of virtual
work, PVW:
T u dS + f u dV =
i
ij
ij
dV
(5.3)
It is interesting to note that the above equation is independent of any constitutive law
and can be applied to all continuous materials within the limitations of small
deformations.
Illustrative Example I
Consider a planar truss as depicted in Fig. 5.2. It is assumed that all the members have
the same cross-sectional area and Youngs modulus.
If we neglect the gravitational effects, the PVW for this problem can be written as:
P u = ij ij dV
V
(5.4)
- 116 -
(2)
8m (1)
10m
30o
A
X1
30o
(3)
X2
10m
1000N
Fig. 5.2
For this problem, we may only consider virtual displacements for the point A. The
changes in the length of each rod due to the imposed virtual displacements u are:
L1 = u1 cos 60 ;
X
L2 = u1 cos 30 ;
X
L3 = u1 cos 30 ;
X1
1
L1 = u2 sin 60
X
L2 = u2 sin 30
X
L3 = u2 sin 30
X2
2
(5.5)
In the above, X Li represent the components of the virtual elongation (or contraction) of
the members in the X directions. Note that, u have been chosen positive in the
positive direction of the axes. Accordingly, we can calculate the components of the
virtual strains as:
cos 60
;
8
cos 30
X1
2 = u1
;
10
cos 30
X1
3 = u1
;
10
1 = u1
X1
sin 60
8
sin 30
X2
2 = u2
10
sin 30
X2
3 = u2
10
X2
1 = u2
(5.6)
where, X i represent the components of the virtual strain of the members in the X
directions.
Now, we will calculate the real stress in each rod, using the real axial strain caused by
the real displacement of the point A.
- 117 -
cos 60
sin 60
cos 60
sin 60
+ u2
1 = Eu1
+ Eu2
8
8
8
8
cos 30
sin 30
cos 30
sin 30
2 = u1
+ u2
2 = Eu1
+ Eu2
10
10
10
10
cos 30
sin 30
cos 30
sin 30
3 = u1
u2
3 = Eu1
Eu2
10
10
10
10
1 = u1
(5.7)
Accordingly, we can apply the PVW in the horizontal direction as follows:

cos 60
cos 60
sin 60
cos 60
P1 u1 = 8 AE u1
u1
+ u2
u1
8
8
8
8
cos 30
cos 30
sin 30
cos 30
+10 AE u1
u1
+ u2
u1
10
10
10
10
(5.8)
cos 30
cos 30
sin 30
cos 30
+10 AE u1
u1
u2
u1
10
10
10
10
which may be simplified to give:

(5.9)
P1
= 0.181u1 + 0.054u2
AE
and also vertically as:
cos 60
sin 60
sin 60
sin 60
P2 u2 = 8 AE u1
u2
+ u2
u2
8
8
8
8
cos 30
sin 30
sin 30
sin 30
+10 AE u1
u2
+ u2
u2
10
10
10
10
(5.10)
cos 30
sin 30
sin 30
sin 30
+10 AE u1
u2
u2
u2
10
10
10
10
which upon simplifying gives the second required equation:

(5.11)
P2
= 0.054u1 + 0.143u2
AE
Now, we may set: P = (0 , 1000) , and solve Eqs. (5.9) and (5.11) to give:
u1 =
2342
AE
u2 =
7846
AE
(5.12)
- 118 -
Principle of Complementary Virtual Work

In order to establish another useful variational principle, we may vary the stress field and
applied forces while keeping the displacement field fixed.
Wvirt
= u. TdS + u. fdV
(5.13)
= ui Ti dS + ui fi dV
S
Using an approach similar to what used for PVW, we may write the Principle of
Complementary Virtual Work, PCVW, as:
u T dS + u f dV =
i
ij
ij
(5.14)
dV
Illustrative Example II
Consider a simple pin-connected structure as depicted in Fig. 5.3. The aim is to
determine the vertical movement of the point C. We assume linear elastic behavior,
identical cross section, and same modulus of elasticity for all the members.
1000N
E
60o
1000
500
612
1433 10m
-866
0.5F
-1412
0.707F
1000N
10m
C
F
1433 10m
-2027
0.5F
Fig. 5.3
-0.707F
- 119 -
Since we require the vertical displacement of the point C, we apply a virtual vertical load
at this point. Next, we will obtain the virtual stresses due to this virtual load.
0.707 F
A
0.707 F
=
A
0.5 F
= BD =
A
CD =
AC
AB
(5.15)
Now we obtain the real strains in the relevant members. Note that we only consider those
members which are affected by the virtual load.
612
AE
2027
=
AE
1433
= BD =
AE
CD =
AC
AB
(5.16)
Accordingly, the principle of complementary virtual work will be:

612 0.707 F
2027 0.707 F
14.14 A +
14.14 A
A
A
AE
AE
1433 0.5 F
+2
10 A
AE A
Fv =
(5.17)
which can be solved for the displacement to give:

v=
40720
AE
(5.18)
Principle of Total Potential Energy

The general form of the constitutive expressions for an elastic continuum can be written
as:
ij =
ij
(5.19)
- 120 -
in which is the density and is the strain energy density. If we substitute the above
expression for stresses in Eq. (5.3), we will have:
T u dS + f u dV =
i
ij dV
ij
Ti ui dS + fi ui dV = dV = dV = U
S
(5.20)
where U is the total strain energy stored in the body. The left side of Eq. (5.20) may be
defined as a variation in the potential energy, V , so we may write:
V = U
(U + V ) = 0
(5.21)
or = 0
in which is the total potential energy of the body.
Strain Energy
For elastic bodies the strain energy can be defined as:
U =
ij
ij d ij dV
(5.22)
If we substitute the constitutive equation for the linear elastic response of isotropic
materials into Eq. (5.22), we have:
U = kk jj + ij ij dV
V 2
(5.23)
for which the one-dimensional form is:

U =
EededV =
1
1
Ee 2 dV = edV
2 V
2 V
Assignment 14:
Solve the illustrative example I, using the principle of total potential energy.
(5.24)
- 121 -
Principle of Total Complementary Energy

The principle of total complementary energy can be derived from the principle of
complementary virtual work by considering a specific complementary strain energy
density as a function of stresses such that:
ij =
(5.25)

ij
Accordingly, we may write:
ui Ti dS + ui fi dV =
S

ij dV
ij
= dV = dV = U
V
(5.26)
where U is the complementary strain energy stored in the body. The left side of Eq.
(5.26) may be defined as a variation in the potential energy, V , so we may write:
V = U
(5.27)
(U + V ) = 0
or = 0
in which is the total complementary energy of the body.
Complementary Strain Energy

For elastic bodies the complementary strain energy can be defined as:
U =
ij
ij d ij dV
(5.28)
If we substitute the constitutive equation for the linear elastic response of isotropic
materials into Eq. (5.28), we will have:
1

U =
kk jj + ij ij dV
V
G
2E
for which the one-dimensional form is:
(5.29)
U =
d dV =
- 122 (5.30)
1 2
1
dv = edV
V
2 E
2 V
It is clear that only for linear elasticity we have U = U .
The First Castigliano Theorem

In this section we will derive the first Castigliano theorem using the principle of total
potential energy. Consider an elastic body maintained in a state of static or dynamic
equilibrium as depicted in Fig. 5.4. The body rests on a system of rigid supports while
being under the action of externally applied loads.
QN
Q2
Fig. 5.4
N
j =1
j =1
= U ( i ) Q j j ; V = Q j j
(5.31)
Implementing the principle of total potential energy, we have:

N
=
i =1
i
= 0
i
i Q i i = 0
i =1
U
Q i = i ; i = 1,K , N
(5.32)
- 123 -
The above equation shows that if the strain energy can be computed as a function of
displacements, the related force or torque for a particular displacement component can be
determined by differentiation.
Illustrative Example III

Consider the problem shown in Fig. 5.5. We intend to determine the force required to
cause the joint to descend a given distance .
Fig. 5.5
The deformed length of each member is:

L + cos
(5.33)

e=
cos
e2
U = 2 dedV = 2 ALE
V 0
2
EA
cos 2 2
=
L
e
(5.34)
Now, using the first castiglianos theorem, we have:

P=
dU 2 AE
=
cos 2
d
L
(5.35)
- 124 -
The Second Castigliano Theorem

Here, we utilize the principle of total complementary energy.
N
j =1
j =1
= U ( Q i ) Q j j ; V = Q j j
(5.36)
Accordingly, we may write:

N
=
i =1

Qi = 0
i
Q

i Q i = 0
i
i =1 Q
U
i =
; i = 1,K , N
Q i
N
(5.37)
The above equation shows that if the complementary strain energy can be computed as a
function of forces, the related displacement for a particular force component can be
determined by differentiation.
An Introduction to Energy Functionals

In general, we may present a partial differential equation by:
Lu = f
(5.38)
where L is an operator acting on the dependent variable u , and f is a function of the

independent variable and is called the driving or forcing function. We may also define
the inner product of two functions, g and h , over the domain of the problem, V, as:
( g, h ) = V g hdV
or
( g , h ) = V ghdV
(5.39)
If the operator L is self-adjoint or symmetric, we will have:
( Lu, v ) = ( u , Lv )
(5.40)
In our discussions we will also require that L be positive definite for all functions u
satisfying homogenous boundary conditions:
( Lu, u ) > 0
(5.41)
- 125 -
The general expression for homogenous boundary condition is:
ui +
ui
=0
n
(5.42)
where and are constants and n is the normal to the boundary. Now we may define
a function I (u ) , called a quadratic functional, such that:
I (u ) =
(5.43)
1
( Lu, u ) (u, f )
2
It can be shown that if the equation Lu = f has a solution, and if L is a self-adjoint

positive definite operator, then the function u that minimizes I (u ) is the solution to
Lu = f . The minimization procedure can also be used to find approximate solutions.
Variational Principles Applied to Beams

We start with the definition of the strain energy for a linear elastic beam:
U =U =
ij
ij d ij dV
1 l h/2
= b 1111dx3 dx1
2 0 h / 2
l h/2
1
= bE 112 dx3 dx1
0 h / 2
2
l h/2
2
1
= bE ( x3,11 ) dx3 dx1
0 h / 2
2
2
EI l
=
,11 ) dx1
(
2 0
(5.44)
The potential V is defined by:

l
V = f i ui dV Ti ui dS = q dx1
V
(5.45)
Finally, the total potential energy can be written as:

l EI
2
= U + V = (,11 ) q dx1
0
2
which has the same form as Eq. (5.43).
(5.46)
- 126 -
The Ritz Technique

Based on the above definitions it may be shown, either by the application of variational
principles to the Navier displacement equations of equilibrium or by comparing Eq.(5.46)
with Eq.(5.43), that for homogenous boundary conditions the quadratic functional is:
I (u ) =
(5.47)
Hence, in order to find approximate solutions to our beam problem, we may apply the
Ritz technique to the total potential energy. The procedure starts with the approximate
displacement field defined as:
n
ui( n ) = ai( j )i( j )
(5.48)
j =1
The ai( j ) are called the Ritz coefficients and i( j ) are known as coordinate functions.
The coefficients ai( j ) are determined so as to minimize the total potential energy ( n ) ,
which implies that:
(5.49)
( n )
= 0 , j = 1,L , n
ai( j )
The above differentiation results in 3n equations for the unknown coefficients.
Illustrative Example IV
Let us find approximate expressions for the deflection of the beam shown in Fig. 5.6.
P
L/2
X1
X3
Fig. 5.6
- 127 -
The exact solution to this problem is:

(5.50)
3
PL3 x1 x1
L
=
3 4 ; x1 <
48 EI L L
2
We start by finding the total potential energy as follows:

(5.51)
L EI
2
= U + V = (,11 ) dx1 P x = L
1
0
2
2
We may choose our coordinate function as:
(1) = a (1) sin
x1
(5.52)
Accordingly the total potential energy can be written as:

(5.53)
(1) =
EI (1) 2 L
a ) Pa (1)
(
2
L 2
Now we may find the coefficients through the minimization of (1) and determine the
displacement function as follows:
4
(1) EIL (1)

a P=0
=
a (1)
2
L
2 PL3
a = 4
EI
x
2 PL3
(1) = 4 sin 1
L
EI
(5.54)
(1)
Finally, we compare the obtained result with the exact solution at two different points:
L
4
L
2
(exact)
0.01432
0.02083
EI (1)
(Ritz)
PL3
0.01452
0.02053
EI
PL3
which shows a very good agreement.

Mirzaei ElasticityLecture PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mirzaei ElasticityLecture PDF

Uploaded by

Copyright:

Available Formats

M.

Timoshenko & Goodier, "Theory of Elasticity"

Tensors, Indicial Notation

Kinematics of Deformable Solids

Suppose a material point is at position X in the undeformed solid, and moves to a

Furthermore, the lines stretch by the same amount, i.e.:

(ds ) 2 = dx12 + dx22 + dx32 = dxi dxi = dx dx

(ds) 2 (dS ) 2 = dxi dxi dX i dX i

and proceed as follows:

Rewriting the indices in a more proper form, we get

Alternatively we may write

(ds ) 2 (dS ) 2 = dxi dxi X i , j dx j X i ,k dxk = dx dx (F dx) (F dx)

Replacing the dummy indices i and k, we have

In practice, we need to make a number of assumptions to simplify the equations of linear

The linear strain tensor in Cylindrical and Spherical

ij ,kl + kl ,ij = ik , jl + jl ,ik

which represents 81 equations, only six of them are independent.

11,22 + 22,11 = 212,12

23,21 + 21,23 31,22 = 22,31

Kinetics of Deformable Solids

in which dA is an element of area on a surface subjected to a force dP (see Fig. 1.6).

in which dV denotes an infinitesimal volume element subjected to a force dP (Fig. 1.7).

Internal forces induced by external loading

Here, we may write:

Tn dAn Ti dAi + f hdAn = 0,

dAi = dAn Cos(n, ei )

which implies that

The characteristic polynomial is

1 ( ij ij )n j = 0 n11 , n12 , n31

Conservation of Linear Momentum

TdA + fdV = u&&dV

which, using (1-28), can be written as

Now we use the Divergence theorem of Gauss and write

The Equations of Motion in Cylindrical and Spherical Coordinate Systems can be

Conservation of Angular Momentum

(X T)dA + (X f )dV = (X u&&)dV

ijk [( X i lj ),l + X i f j ]dV = 0

Next, we use (1-41) to simplify the above as:

Since we are dealing with an arbitrary volume, we can write:

which shows that the stress tensor is symmetric.

Constitutive Equations of Linear Elasticity

which can be simplified to:

Similarly, we can show that:

Hence, we may write:

which results in:

We may have an additional simplification by considering the directional independence in

Formulation and Solution Methods of Elasticity Problems

along with the compatibility and stress boundary equations:

ij ,kl + kl ,ij = ik , jl + jl ,ik

Classical Displacement Formulation

ij uk ,ki + (ui , ji + u j ,ii ) + f j = 0

Finally, using i for the dummy index in all terms we write:

u j ,ii + ( + )ui ,ij + f j = 0

Navier Displacement Equations in Cylindrical and Spherical

Classical Force Formulation

ij ,kl + kl ,ij ik , jl jl ,ik =

However, only six of the above equations are independent:

ij ,kk + kk ,ij ik , jk jk ,ik =

Beltrami-Michell Equations in Cylindrical Coordinates:

Beltrami-Michell Equations in Spherical Coordinates:

Chapter 2: Some Representative Boundary Value

On the top surface we have:

Ti (0, 0, 0), ni (n1 , n2 , 0)