Documente Academic
Documente Profesional
Documente Cultură
\
|
=
+ =
u (1.1.2)
Cylindrical coordinates are used mostly for volume and surface area
analysis of revolved solids. Examples of resolved solids are cone sections,
hyperboloids of one sheet, sphere sections, elliptical and circular cylinders.
We now pursue the development of the strain-displacement relations
(Cauchy's relations) in cylindrical coordinates. Starting with the form
| |, ) (
2
1
T
u u V + V = (1.2.1)
02
where is the strain matrix and u V is the displacement gradient matrix
and
T
u) (V is its transpose. The displacement vector and strain tensor can
be expressed by
,
z z r r
e u e u e u u + + =
u u
,
(
(
(
=
z z zr
z r
rz r r
c c c
c c c
c c c
u
u u u
u
(1.2.2)
where
i
e are the unit basis vectors in the curvilinear system.
The derivative operation in cylindrical coordinates can be expressed by
,
1
z r
e
z
u
e
u
r
e
r
u
u
c
c
+
c
c
+
c
c
= V
u
u
z r
z
r r r
r
e e
r
u
e e
r
u
e e
r
u
c
c
+
c
c
+
c
c
=
u
u
z
z
r r
r
e e
u
r
e e
u
u
r
e e u
u
r
u u u
u
u u
u u u c
c
+
|
.
|
\
|
c
c
+ +
|
.
|
\
|
c
c
+
1 1 1
.
z z
z
z r z
r
e e
z
u
e e
z
u
e e
z
u
c
c
+
c
c
+
c
c
+
u
u
(1.2.3)
Placing this result into the strain-displacement from, Eq. (1.2.3), gives the
desired relations in cylindrical coordinates. The individual scalar equations
are given by
,
r
u
r
r
c
c
= c ,
1
|
.
|
\
|
c
c
+ =
u
c
u
u
u
u
r
r
,
z
u
z
z
c
c
= c
,
1
2
1
|
.
|
\
|
c
c
+
c
c
=
r
u
r
u u
r
r
r
u u
u
u
c
,
1
2
1
|
.
|
\
|
c
c
+
c
c
=
u
c
u
u
z
z
u
r z
u
.
2
1
|
.
|
\
|
c
c
+
c
c
=
r
u
z
u
z r
zr
c (1.2.4)
07
As mentioned in the introduction, in order to solve many elasticity
problems, formulation must be done in curvilinear coordinates typically
using cylindrical or spherical systems. Thus by following similar methods
as used with strain-displacement relations, we now wish to develop
expressions for the equilibrium equations in curvilinear cylindrical
coordinates. By using a direct vector/matrix notation, the equilibrium
equations can be expressed as
. 0 = + V F (1.3.1)
where
j i ij
e e o = is the stress matrix or dyadic, and F is the body force
vector. The desired curvilinear expression can be obtained from Eq. (1.3.1)
by using the appropriate form for . V
Figure 1.2: Stress components in cylindrical coordinates.
00
Cylindrical coordinates were originally presented in Figure 1.1. For
such a system, the stress components are defined on the differential
element shown in Figure 1.2, and thus the stress matrix is given by
.
(
(
(
=
z z zr
z r
rz r r
o t t
t o t
t t o
u
u u u
u
(1.3.2)
Now, the stress can be expressed in terms of traction components as
,
z z r r
e T e T e T + + =
u u
(1.3.3)
where
.
,
,
z z z r zr z
z z r r
z rz r r r r
e e e T
e e e T
e e e T
o t t
t o t
t t o
u u
u u u u u
u u
+ + =
+ + =
+ + =
(1.3.4)
The divergence operation in the equilibrium equations can be written as
,
1 1
z
T T
r
T
r r
T
z
r
r
c
c
+
c
c
+ +
c
c
= V
u
u
) (
1
z rz r r r z
rz r
r
r
e e e
r
e
r
e
r
e
r
t t o
t t o
u u u
u
+ + +
c
c
+
c
c
+
c
c
=
|
.
|
\
|
c
c
+
c
c
+ +
c
c
+
z
z
r r r
r
e e e e e
r u
t
o
u
o
t
u
t
u
u u
u
u u
u
1
.
z
z z
r
rz
e
z
e
z
e
z c
c
+
c
c
+
c
c
+
o t t
u
u
(1.3.5)
Combining this result into Eq. (1.3.1) gives the vector equilibrium equation
in cylindrical coordinates. The three scalar equations expressing
equilibrium in each coordinate direction then became
, 0 ) (
1 1
= + +
c
c
+
c
c
+
c
c
r r
rz r r
F
r z r r
u
u
o o
t
u
t o
02
, 0
2 1
= + +
c
c
+
c
c
+
c
c
u u
u u u
t
t
u
o t
F
r z r r
r
z r
. 0
1 1
= + +
c
c
+
c
c
+
c
c
z rz
z z rz
F
r z r r
t
o
u
t t
u
(1.3.6)
It is interesting to note that the equilibrium equations in cylindrical
coordinates contain additional terms not involving derivatives of stress
components. The appearance of these terms can be explained
mathematically because of the curvature of the space. However, a more
physical interpretation can be found by redeveloping these equations
through a simple force balance analysis on the appropriate differential
element. In general, Eqs. (1.3.6) look much more complicated when
compared to the Cartesian form. However, under particular conditions, the
curvilinear forms lead to an analytical solution that could not be reached by
using Cartesian coordinates.
( )
In order to construct a general three dimensional constitutive law for
linear elastic materials, we assume that each stress component is linearly
related to each strain component
, 2 2 2
16 15 14 13 12 11 zr z r z r r
C C C C C C c c c c c c o
u u u
+ + + + + =
, 2 2 2
26 25 24 23 22 21 zr z r z r
C C C C C C c c c c c c o
u u u u
+ + + + + =
, 2 2 2
36 35 34 33 32 31 zr z r z r z
C C C C C C c c c c c c o
u u u
+ + + + + =
, 2 2 2
46 45 44 43 42 41 zr z r z r r
C C C C C C c c c c c c t
u u u u
+ + + + + =
, 2 2 2
56 55 54 53 52 51 zr z r z r z
C C C C C C c c c c c c t
u u u u
+ + + + + =
, 2 2 2
66 65 64 63 62 61 zr z r z r zr
C C C C C C c c c c c c t
u u u
+ + + + + = (1.4.1)
02
where the coefficients
ij
C are material parameters and the factors of 2 arise
because of the symmetry of the strain. Note that this relation could be
expressed by writing the strains as a linear function of the stress
components. These relations can be cast into a matrix format as
.
2
2
2
... ... ... ...
... ... ... ... ... ...
... ... ... ... ... ...
... ... ... ... ... ...
... ... ... ... ...
... ... ...
66 61
21
16 12 11
(
(
(
(
(
(
(
zr
z
r
z
r
zr
z
r
z
r
C C
C
C C C
c
c
c
c
c
c
t
t
t
o
o
o
u
u
u
u
u
u
(1.4.2)
The above relations can also be expressed in standard tensor notation by
writing
,
kl ijkl ij
C c o = (1.4.3)
where
ijkl
C is a fourth-order elasticity tensor whose components include
all the material parameters necessary to characterize the material. Based on
the symmetry of the stress and strain tensors, the elasticity tensor must
have the following properties:
.
jikl ijkl
C C = (1.4.4)
In general, the fourth-order tensor
ijkl
C has 81 components. However, Eq.
(1.4.4) reduces the number of independent components to 36, and this
provides the required match with form of Eqs. (1.4.1) or (1.4.2). The
components of
ijkl
C or equivalently
ij
C are called elastic moduli and have
units of stress (force/area). In order to continue further, we must address
the issues of material homogeneity and isotropy.
If the material is homogenous, the elastic behavior dose not vary
spatially, and thus all elastic moduli are constant. For this case, the
02
elasticity formulation is straightforward, leading to the development of
many analytical solutions to problems of engineering interest. A
homogenous assumption is an appropriate model for most structural
applications, and thus we primarily choose this particular case for
subsequent formulation and problem solution.
Similar to homogeneity, another fundamental material is isotropy. This
property has to do with differences in material moduli with respect to
orientation. For example, many materials including crystalline minerals,
wood, and fiber-reinforced composites have different elastic moduli in
different directions. Materials such as these are said to be anisotropic. Note
that for most real anisotropic materials there exist particular directions
where the properties are the same. These directions indicate material
symmetries. However, for many engineering materials (most structural
metals and many plastics), the orientation of crystalline and grain
microstructure is distributed randomly so that macroscopic elastic
properties are found to be essentially the same in all directions. Such
materials with complete symmetry are called isotropic. As expected, an
anisotropic model complicates the formulation and solution of problems.
The tensorial form of Eq. (1.4.3) provides a convenient way to establish
the desired isotropic stress-strain relations. If we assume isotropic
behavior, the elasticity tensor must be the same under rotating of the
coordinate system. The most general form that satisfies this isotropy
condition is given by
,
jk il jl ik kl ij ijkl
C o o o |o o oo + + = (1.4.5)
where o , | and are arbitrary constants.
Using the general form of Eq. (1.4.5) in stress-strain relation, Eq.
(1.4.3), gives
02
, 2
ij ij kk ij
c o c o + = (1.4.6)
where we have relabeled particular constants using and . The elastic
constant is called lam's constant and is referred to as the shear
modulus or modulus of rigidity.
Eq. (1.4.6) can be written out in individual scalar equations as
, 2 ) (
r z r r
c c c c o
u
+ + + =
, 2 ) (
u u u
c c c c o + + + =
z r
, 2 ) (
z z r z
c c c c o
u
+ + + =
, 2
u u
c t
r r
=
, 2
z z u u
c t =
. 2
zr zr
c t = (1.4.7)
Relations given in Eqs. (1.4.6) or (1.4.7) are called the generalized
Hooke's law of linear isotropic elastic solids. They are named after Robert
Hooke who in 1678 first proposed that the deformation of an elastic
structure is proportional to the applied force. Notice the significant
simplicity of the isotropic form when compared to the general stress-strain
low originally given by Eq. (1.4.1). It should be noted that only two
independent elastic constants are needed to describe the behavior or
isotropic materials.
Stress-strain relations of Eqs. (1.4.6) or (1.4.7) may be inverted to
express the strain in terms of the stress. In order to do this, it is convenient
to use the index notation from Eq. (1.4.6) and set the two free indices the
same to get
. ) 2 3 (
kk kk
c o + = (1.4.8)
01
This relation can be solved for
kk
c and substituted back into Eq. (1.4.6) to
get
,
2 2 2
1
|
.
|
\
|
+
=
ij kk ij ij
o o
o
c (1.4.9)
which is more commonly written as
,
1
ij kk ij ij
E E
o o
v
o
v
c
+
= (1.4.10)
where
,
) 2 3 (
+
+
= E (1.4.11)
is called the modulus of elasticity or Young's modulus, and
,
) ( 2
v
+
= (1.4.12)
is referred to as Poisson's ratio. The index notation relation of Eq. (1.4.10)
may be written out in component (scalar) form giving the six equations
| |, ) (
1
z r r
E
o o v o c
u
+ =
| |, ) (
1
r z
E
o o v o c
u u
+ =
| |, ) (
1
u
o o v o c + =
r z z
E
,
2
1 1
u u u
t
t
v
c
r r r
E
=
+
=
,
2
1 1
z z z
E
u u u
t
t
v
c =
+
=
.
2
1 1
zr zr zr
E
t
t
v
c =
+
= (1.4.13)
01
Constitutive form of Eqs. (1.4.10) or (1.4.13) again illustrates that only two
elastic constants are needed to formulate Hooke's law for isotropic
materials. By using any of the isotropic forms of Hooke's law, it can be
shown that the principal axes of stress coincide with the principal axes of
strain. This result also holds for same but not all anisotropic materials.
We now wish to develop the reduced set of field equations solely in
terms of the displacements. This system is referred to as the displacement
formulation and is most useful when combined with displacement-only
boundary conditions. This is easily accomplished by using the strain-
displacement relations in Hooke's law to give
), (
, , , i j j i ij k k ij
u u u + + = o o (1.5.1)
which can be expressed as six scalar equations
, 2
1
) (
1
r
u
z
u u
r
ru
r r
r z
r r
c
c
+
|
.
|
\
|
c
c
+
c
c
+
c
c
=
u
o
u
,
1
2
1
) (
1
|
.
|
\
|
c
c
+ +
|
.
|
\
|
c
c
+
c
c
+
c
c
=
u
u
o
u u
u
u
u
r z
u u
r
ru
r r
r
z
r
, 2
1
) (
1
z
u
z
u u
r
ru
r r
z z
r z
c
c
+
|
.
|
\
|
c
c
+
c
c
+
c
c
=
u
o
u
,
1
|
.
|
\
|
c
c
+
c
c
=
r
u
r
u u
r
r
r
u u
u
u
t
,
1
|
.
|
\
|
c
c
+
c
c
=
u
t
u
u
z
z
u
r z
u
.
|
.
|
\
|
c
c
+
c
c
=
r
u
z
u
z r
zr
t (1.5.2)
01
Using these relations in the equilibrium equations gives the result
, 0 ) (
, ,
= + + +
i ki k kk i
F u u (1.5.3)
which are the equilibrium equations in terms of the displacements and are
referred to as Navier's or Lam's equations. This system can be expressed
in vector form as
, 0 ) ( ) (
2
= + V V + + V F u u (1.5.4)
or written out in terms of the three scalar equations
|
.
|
\
|
c
c
+
c
c
+
c
c
c
c
+ + |
.
|
\
|
c
c
V
z
u u
r
ru
r r r
u
r r
u
u
z
r
r
r
u
u
u u
1
) (
1
) (
2
2 2
2
, 0 = +
r
F (1.5.5)
|
.
|
\
|
c
c
+
c
c
+
c
c
c
c
+ +
|
|
.
|
\
|
c
c
+ V
z
u u
r
ru
r r r
u
r r
u
u
z
r
r
u u
u
u u
u
1
) (
1 1
) (
2
2 2
2
, 0 = +
u
F (1.5.6)
, 0
1
) (
1
) (
2
= +
|
.
|
\
|
c
c
+
c
c
+
c
c
c
c
+ + V
z
z
r z
F
z
u u
r
ru
r r z
u
u
u
(1.5.7)
where the Laplacian is given by
.
1 1
2
2
2
2
2
2
z r
r
r
r r
c
c
+
c
c
+
|
.
|
\
|
c
c
c
c
= V
u
(1.5.8)
Navier's equations are the desired formulation for the displacement
problem, and the system represents three equations for the three unknown
displacement components. This system is still difficult to solve, and
additional mathematical techniques have been developed to further
simplify these equations for problem solution. Common methods normally
employ the use of displacement potential functions.
03
The purpose of this chapter is to state some general forms of
differential equations and some kinds of boundary-value
problems (BVPs). In addition, this chapter states some numerical
solutions. The FDM as well as the modified R-K method are
presented. In addition, the least square and Richardson
extrapolation are also presented.
To obtain accurate numerical solutions to differential equations
governing physical systems has always been an important problem with
scientists and engineers. These differential equations basically fall into two
classes, ordinary and partial, depending on the number of independent
variables present the differential equations: one for ordinary and more than
one for partial (see Jain [92]).
The general form of the ordinary differential equation can be written as
, ] [ g L = (2.1.1)
where L is a differential operator and g is a given function of the
independent variable r. The order of the differential equation is the order of
its highest derivative and its degree is the degree of the derivative of the
highest order after the equation has been rationalized. If no product of the
03
dependent variable ) (r with itself or any of its derivatives occur, the
equation is said to be linear, otherwise it is nonlinear. A linear differential
equation of order m can be expressed in the form
), ( ) ( ) ( ] [
0
) (
r g r r f L
m
p
p
p
= =
=
(2.1.2)
in which ) (r f
p
are known functions. The general nonlinear differential
equation of order m can be written as
( ) , 0 , ,..., , ,
) ( ) 1 (
= '
m m
r F (2.1.3)
or
( ), ,..., , , ) (
) 1 ( ) (
' =
m m
r f r (2.1.4)
which is called a canonical representation of differential equation given in
Eq. (2.1.3). In such a form, the highest order derivative is expressed in
terms of the lower order derivatives and the independent variable. The
general solution of the mth order ordinary differential equation contains m
independent arbitrary constants. In order to determine the arbitrary
constants in the general solution if the m conditions are prescribed at one
point, these are called initial conditions. The differential equation together
with the initial conditions is called the initial-value problem. Thus, the mth
order can be expressed as
( )
. 1 ,..., 2 , 1 , 0 , ) (
, ,..., , , ) (
) (
0
0
) (
) 1 ( ) (
= =
' =
m p r
r f r
p p
m m
(2.1.5)
If the m conditions are presented at more than one point, these are
called boundary conditions. The differential equation together with the
boundary conditions is known as the boundary-value problem.
03
A general boundary-value problem can be represented symbolically as
, , ,... 2 , 1 , ] [
, ] [
m g U
g L
= =
=
,
, ,
(2.2.1)
where L is an mth order differential operator, g is a given function and
,
U
are the boundary conditions. We shall use r as an independent variable for
the boundary-value problem.
If L represents an mth order linear differential operator and ] [
,
U
represent two point boundary conditions, then Eq. (2.2.1) can be expressed
in the form
) (
0
) ( ] [
v
m
v
v
r f L
=
=
], , [ ), ( ) ( ... ) ( ) (
) (
1 0
b a r r g r f r f r f
m
m
e = + + ' + = (2.2.2)
( ) . ,..., 2 , 1 , ) ( ) ( ] [
1
0
) (
,
) (
,
m b b a a U
m
k
k
k
k
k
= = + =
=
,
, , , ,
(2.2.3)
For m=2q, the k boundary conditions which are linearly independent
and contain only derivatives up to ) 1 ( q th order are called the essential
boundary conditions, and the remaining ) 2 ( k q boundary conditions are
termed the suppressible boundary conditions.
The simplest boundary-value problem is given by second order
differential equation
], , [ ), ( ) ( ) ( ) (
0 1 2
b a r r g r f r f r f e = + ' + ' ' (2.2.4)
with one of the three boundary conditions given below.
The boundary conditions of the first kind are:
00
(i)
1
) ( = a and . ) (
2
= b
The boundary conditions of the second kind are:
(ii)
1
) ( = ' a and . ) (
2
= ' b
The boundary conditions of the third kind, sometimes called Sturm's
boundary conditions, are:
(iii)
1 1 0
) ( ) ( = ' a a a a and . ) ( ) (
2 1 0
= + ' b b b b
where
1 0 0
, , a b a and
1
b are all positive constants.
In Eq. (2.2.1) if , 0 ) ( = r g the differential equation is called
homogeneous; otherwise it is inhomogeneous. Similarly, the boundary
conditions are called homogeneous when
,
are zero; otherwise
inhomogeneous. The boundary-value problem is called homogeneous if the
differential equation and the boundary conditions are homogeneous. A
homogeneous boundary-value problem ) 0 , 0 ) ( ( = =
,
r g possesses only a
trivial solution . 0 ) ( = r We, therefore, consider those boundary-value
problem in which a parameter q occurs either in differential equation or in
the boundary conditions, and we determine values of q , called eigenvalues,
for which the boundary-value problem has a nontrivial solution. Such a
solution is called eigen-function and the entire problem is called an
eigenvalues or characteristic value problem.
In the boundary-value problems, the arbitrary constants in the solution
are determined from the conditions given at more than one point.
Therefore, it is possible for more than one solution to exist or no solution
may exist.
In general, a boundary-value problem does not always have a unique
solution. However, the existence and uniqueness of the solution for a
03
special class of boundary-value problems, called class M, can be
established.
A boundary-value problem will be called of class M if it is of the form
, ) ( , ) ( ), , (
2 1
= = = ' ' b a r f (2.2.5)
and,
(i) the initial-value problem
, ) ( , ) ( ), , (
1
A a a r f = ' = = ' ' (2.2.6)
with A arbitrary, has a unique solution, and ) , ( r f is such that
(ii) ) , (
r f is continuous
0 ) , ( >
r f for ) , ( ], , [ e e b a r . (2.2.7)
In what follow we will discuss some boundary-value problems and we
will investigate their analytical and numerical solutions.
The finite difference method (FDM) was first developed by A. Thom in
the 1920s under the title the method of square to solve nonlinear
hydrodynamic equations (Thom and Apelt [93]). The finite difference
techniques are based upon the approximations that permit replacing
differential equations by finite difference equations. These finite difference
approximations are algebraic in form, and the solutions are related to grid
points (see Linz and Wang [94]).
Thus, a finite difference solution basically involves three steps:
1. Dividing the solution into grids of nodes.
03
2. Approximating the given differential equation by finite difference
equivalence that relates the solutions to grid points.
3. Solving the difference equations subject to the prescribed boundary
conditions and/or initial conditions.
Figure 2.1: Common two-dimensional grid patterns
FDM scheme:
Differential equations
estimating derivatives numerically
finite difference equations.
The error in a method's solution is defined as the difference between its
approximation and the exact analytical solution. The two sources of error
in finite difference methods are round-off error, the loss of precision due to
computer rounding of decimal quantities, and truncation error or
discretization error, the difference between the exact solution of the finite
03
difference equation and the exact quantity assuming perfect arithmetic
(that is, assuming no round-off).
Figure 2.2: The finite difference method
To use a finite difference method to attempt to solve (or, more
generally, approximate the solution to) a problem, one must first discretize
the problem's domain. This is usually done by dividing the domain into a
uniform grid (see Figure 2.2). Note that this means that finite-difference
methods produce sets of discrete numerical approximations to the
derivative, often in a time-stepping manner.
An expression of general interest is the local truncation error of a
method. Typically expressed using Big-O notation, local truncation error
refers to the error from a single application of a method. That is, it is the
quantity
i i
f R f ) ( if ) (
i
R f refers to the exact value and
i
f to the
numerical approximation. The remainder term of a Taylor polynomial
... ) (
! 2
) (
! 1
) (
) ( ) (
2
0 0
0 0
+ A
' '
+ A
'
+ = A + R
R f
R
R f
R f R R f
), ( ) (
!
) (
0
) (
R R R
n
R f
n
n
n
+ A + (2.3.1)
03
is convenient for analyzing the local truncation error. Using the Lagrange
form of the remainder
n
R from the Taylor polynomial for ) (
0
R R f A +
which is
( )
. ,
)! 1 (
) (
0 0
1
) 1 (
0
R R R R
n
f
R R R
n
n
n
A + < < A
+
= A +
+
+
(2.3.2)
The dominant term of the local truncation error can be discovered. For
example, again using the forward-difference formula for the first
derivative, knowing that ), ( ) (
0
R i R f R f
i
A + =
, ) (
! 2
) (
) ( ) ( ) (
2
0 0 0
R i
f
R i R f R f R i R f A
' '
+ A ' + = A +
(2.3.3)
and with some algebraic manipulation, this leads to
,
! 2
) (
) (
) ( ) (
0
0 0
R i
f
R f
R i
R f R i R f
A
' '
+ ' =
A
A +
(2.3.4)
and further noting that the quantity on the left is the approximation from
the finite difference method and that the quantity on the right is the exact
quantity of interest plus a remainder, clearly that remainder is the local
truncation error. A final expression of this example and its order is:
). ( ) (
) ( ) (
0
0 0
R O R f
R i
R f R i R f
A + ' =
A
A +
(2.3.5)
This means that, in this case, the local truncation error is proportional to
the step size R A . Ignoring the ) ( R O A term, we get
.
) ( ) (
) (
0 0
0
R i
R f R i R f
R f
A
A +
~ ' (2.3.6)
The expression on the right side of Eq. (2.3.6) is the forward difference
approximation to f ' at
0
R . It is computed using only values of f and
converges to the derivative with order one. In an almost identical way we
get the backward difference
03
.
) ( ) (
) (
0 0
0
R i
R i R f R f
R f
A
A
~ ' (2.3.7)
The forward and backward difference rules have relatively low accuracy,
but it is not hard to get better approximations. Again, from Taylor's
Theorem we know that
), ) (( ) (
2
) (
) )( ( ) ( ) (
3 2
0
0 0 0
R O R i
R f
R i R f R f R i R f A + A
' '
+ A ' + = A +
(2.3.8)
), ) (( ) (
2
) (
) )( ( ) ( ) (
3 2
0
0 0 0
R O R i
R f
R i R f R f R i R f A + A
' '
+ A ' = A
(2.3.9)
so that, neglecting the higher order terms,
.
) ( 2
) ( ) (
) (
0 0
0
R i
R i R f R i R f
R f
A
A A +
~ ' (2.3.10)
This is the centered difference approximation to the first derivative. It
has second order convergence and normally gives a much more accurate
answer than either Eq. (2.3.6) or Eq. (2.3.7).
Many other formulas can be developed along these lines. For example,
the expression
.
) ( 2
) ( 3 ) ( 4 ) 2 (
) (
0 0 0
0
R i
R f R i R f R i R f
R f
A
A + + A +
~ ' (2.3.11)
is another forward difference approximation for the first derivative.
We can also derive approximations for higher derivatives this way.
Taking more terms in the Taylor expansions
3
0
2
0
0 0 0
) (
6
) (
) (
2
) (
) )( ( ) ( ) ( R i
R f
R i
R f
R i R f R f R i R f A
' ' '
+ A
' '
+ A ' + = A +
), ) ((
4
R O A + (2.3.12)
03
3
0
2
0
0 0 0
) (
6
) (
) (
2
) (
) )( ( ) ( ) ( R i
R f
R i
R f
R i R f R f R i R f A
' ' '
A
' '
+ A ' = A
), ) ((
4
R O A + (2.3.13)
we find that
). ) ((
) (
) ( ) ( 2 ) (
) (
2
2
0 0 0
0
R O
R i
R i R f R f R i R f
R f A +
A
A + A +
= ' ' (2.3.14)
Thus,
.
) (
) ( ) ( 2 ) (
) (
2
0 0 0
0
R i
R i R f R f R i R f
R f
A
A + A +
~ ' ' (2.3.15)
This is a very useful centered difference approximation to the second
derivative that has second order accuracy.
Consider a general second order equation (see Jain [92])
( ), , , U U R f U ' = ' ' | |. ,
0
b R Re (2.3.16)
with the initial conditions
, ) (
0 0
U R U =
. 0 0
) ( U R U ' = ' (2.3.17)
We define
( ), , ,
! 2
) (
2
1 i i i
U U R f
R
K '
A
=
, , ,
! 2
) (
1
21
1 21 2 2
2
2
|
.
|
\
|
A
+ ' + ' A + A +
A
= K
R
b
U K a U R a U R a R f
R
K
i i i i
,
2 2 1 1 1
K W K W U R U U
i i i
+ + ' A + =
+
). (
1
2 2 1 1 1
K W K W
R
U U
i i
' + '
A
+ ' = '
+
(2.3.18)
33
where W b a a , , ,
21 21 2
and W' are arbitrary constants to be determined.
The Taylor series expansion gives
...,
! 4
) (
! 3
) (
! 2
) (
4 3 2
1
+
A
+ ' ' '
A
+ ' '
A
+ ' A + =
+
mv
i i i i i i
U
R
U
R
U
R
U R U U
.....
! 3
) (
! 2
) (
3 2
1
+
A
+ ' ' '
A
+ ' ' A + ' = '
+
mv
i i i i i
U
R
U
R
U R U U (2.3.19)
where
)), ( ), ( , (
i i i i
R U R U R f U ' = ' '
, ) (
i U U R i
ff f U f U
'
+ ' + = ' ' '
U R U U U R U U U U RR
mv
i
ff ff U f U f f f U f U
' ' ' '
+ ' + ' + + ' + = 2 2 2 [
2 2
. ] ) (
i U U U R U
ff ff f U f f + + ' + +
' '
(2.3.20)
We may write Eq. (2.3.20) as
,
i i
f U = ' '
,
i i
Df U = ' ' '
,
2
U i i U i
mv
i
f f Df f f D U + + =
'
where
.
U
f
U
U
R
D
i i
' c
c
+
c
c
' +
c
c
=
Equation (2.3.19) becomes
(
i i i i i i
f D
R
Df
R
f
R
U R U U
2
4 3 2
1
! 4
) (
! 3
) (
! 2
) ( A
+
A
+
A
+ ' A + =
+
) ..., + + +
' U i i U
f f Df f
33
( ) ...
! 3
) (
! 2
) (
2
3 2
1
+ + +
A
+
A
+ A + ' = '
' + U i i U i i i i i
f f Df f f D
R
Df
R
Rf U U
(2.3.21)
Simplifying
2
K , we get
|
.
|
\
|
+ ' + A + =
A
' U i U i R i
f f b f U a f a R f K
R
21 2 2 2
2
2
1
) (
2
\
|
' + + ' +
A
+
' ' U R i U U i U U i RR
f U a f f b f U a f a
R
2 2 2
21
4
1
2 2
2
2
2
2
21
2
! 2
) (
), ) ((
3
21 21 2 21 2
R O f f a f U f b a f f b a
U i U U i i U R i
A +
|
.
|
+ ' + +
' '
or
( )
U i i i i
f f a f D a
R
Df a
R
f
R
K
21
2 2
2
4
2
3 2
2
4
) (
2
) (
2
) (
+
A
+
A
+
A
=
), ) ((
5
R O A + (2.3.22)
where we have used
.
2
1
21 2
b a = (2.3.23)
The substitution of
1
K and
2
K in Eq. (2.3.18) yields
i i i i i
Df W a
R
f W W
R
U R U U
2 2
3
2 1
2
1
2
) (
) (
2
) ( A
+ +
A
+ ' A + =
+
), ) (( ) (
4
) (
5
21 2
2 2
2 2
4
R O f f a W f D a W
R
U i i
A + +
A
+
i i i i
Df W a
R
f W W R U U
2 2
2
2 1 1
2
) (
) (
2
1
'
A
+ ' + ' A + ' = '
+
). ) (( ) (
4
) (
4
21 2
2 2
2 2
3
R O f f a W f D a W
R
U i i
A + ' + '
A
+ (2.3.24)
33
On comparing Eq. (2.3.24) with Eq. (2.3.21), we obtain
, 1
2 1
= +W W , 2
2 1
= ' + ' W W
,
3
1
2 2
= W a . 1
2 2
= ' W a (2.3.25)
The coefficients of
4
) ( R A in
1 + i
U and of
3
) ( R A in
1 +
'
i
U of equation
(2.3.24) will not match with the corresponding coefficients in Eq. (2.3.21)
for any choice of
2
a ,
21
a ,
2
W and
2
W' .Thus the local truncation error is
) ) ((
4
R O A in U and ) ) ((
3
R O A in U' . A simple solution of Eq. (2.3.23)
and Eq. (2.3.25) may be written as
,
2
1
2 1
= =W W
,
3
2
21 2
= = a a ,
3
4
21
= b
,
2
1
1
= ' W
.
2
3
2
= ' W
Thus the Runge-Kutta method Eq. (2.3.18) becomes
( ), , ,
! 2
) (
2
1 i i i
U U R f
R
K '
A
=
,
3
4
,
3
2
3
2
,
3
2
! 2
) (
1 1
2
2
|
.
|
\
|
A
+ ' + ' A + A +
A
= K
R
U K U R U R R f
R
K
i i i i
( ).
2
1
2 1 1
K K U R U U
i i i
+ + ' A + =
+
( ). 3
2
1
2 1 1
K K
R
U U
i i
+
A
+ ' = '
+
(2.3.26)
30
The Runge-Kutta method using four K's is given by
( ), , ,
2
) (
2
1 i i i
U U R f
R
K '
A
=
,
1
,
4
1
2
1
,
2 2
) (
1 1
2
2
|
.
|
\
|
A
+ ' + ' A +
A
+
A
= K
R
U K U R U
R
R f
R
K
i i i i
,
1
,
4
1
2
1
,
2 2
) (
2 1
2
3
|
.
|
\
|
A
+ ' + ' A +
A
+
A
= K
R
U K U R U
R
R f
R
K
i i i i
,
2
, ,
2
) (
3 3
2
4
|
.
|
\
|
A
+ ' + ' A + A +
A
= K
R
U K U R U R R f
R
K
i i i i
( ),
3
1
3 2 1 1
K K K U R U U
i i i
+ + + ' A + =
+
( ). 2 2
3
1
4 3 2 1 1
K K K K
R
U U
i i
+ + +
A
+ ' = '
+
(2.3.27)
A mathematical procedure for finding the best-fitting curve to a given
set of points by minimizing the sum of the squares of the offsets (the
residuals) of the points from the curve. The sum of the squares of the
offsets is used instead of the offset absolute values because this allows the
residuals to be treated as a continuous differentiable quantity. However,
because squares of the offsets are used, outlying points can have a
disproportionate effect on the fit, a property that may or may not be
desirable depending on the problem at hand.
33
Figure 2.3: Least squares fitting
Figure 2.4: Least squares offsets
In practice, the vertical offsets from a line (polynomial, surface,
hyperplane, etc.) are almost always minimized instead of the perpendicular
offsets. This provides a fitting function for the independent variable that
estimates u for a given (most often what an experimenter wants),
allows uncertainties of the data points along the - and u-axes to be
incorporated simply, and also provides a much simpler analytic form for
the fitting parameters than would be obtained using a fit based on
33
perpendicular offsets. In addition, the fitting technique can be easily
generalized from a best-fit line to a best-fit polynomial when sums of
vertical distances are used. In any case, for a reasonable number of noisy
data points, the difference between vertical and perpendicular fits is quite
small.
The least square method is a very popular technique used to compute
estimations of parameters and to fit data. It is one of the oldest techniques
of modern statistics as it was first published in 1805 by the French
mathematician Legendre in a now classic memoir. Nowadays, the least
square method is widely used to find or estimate the numerical values of
the parameters to fit a function to a set of data and to characterize the
statistical properties of estimates. It exists with several variations: Its
simpler version is called ordinary least squares (OLS), a more sophisticated
version is called weighted least squares (WLS), which often performs
better than OLS because it can modulate the importance of each
observation in the final solution. Recent variations of the least square
method are alternating least squares (ALS) and partial least squares (PLS).
The linear least squares fitting technique is the simplest and most
commonly applied form of linear regression and provides a solution to the
problem of finding the best fitting straight line through a set of points. In
fact, if the functional relationship between the two quantities being graphed
is known to within additive or multiplicative constants, it is common
practice to transform the data in such a way that the resulting line is a
straight line, say by plotting T vs instead of T vs in the case of
analyzing the period T of a pendulum as a function of its length . For this
reason, standard forms for exponential, logarithmic, and power laws are
often explicitly computed. The formulas for linear least squares fitting
were independently derived by Gauss and Legendre. Vertical least squares
33
fitting proceeds by finding the sum of the squares of the vertical deviations
Eof a set of n data points
| | , ) ,...., , , (
2
2 1
u =
N i i
a a a f E (2.4.1)
from a function f . Note that this procedure does not minimize the actual
deviations from the line (which would be measured perpendicular to the
given function). In addition, although the unsquared sum of distances
might seem a more appropriate quantity to minimize, use of the absolute
value results in discontinuous derivatives which cannot be treated
analytically. The square deviations from each point are therefore summed,
and the resulting residual is then minimized to find the best fit line. This
procedure results in outlying points being given disproportionately large
weighting.
The condition for E to be a minimum is that
, 0 =
c
c
i
a
E
, 0 =
c
c
i
b
E
(2.4.2)
for N i ..., , 2 , 1 = . For a linear fit,
, ) , ( b a b a f + = (2.4.3)
so
| | , ) ( ) , (
1
2
=
+ u
N
i
i i
b a b a E (2.4.4)
| | , ) 1 ( ) ( 2
1
=
+ u =
c
c
N
i
i i
b a
a
E
(2.4.5)
| | . ) ( ) ( 2
1
=
+ u =
c
c
N
i
i i i
b a
b
E
(2.4.6)
33
Setting 0 =
c
c
=
c
c
b
E
a
E
, and dividing by 2 yields
,
1 1
= =
u = +
N
i
i
N
i
i
b a N (2.4.7)
.
1 1
2
1
= = =
u = +
N
i
i i
N
i
i
N
i
i
b a (2.4.8)
In matrix form,
,
1
1
1
2
1
1
(
(
(
(
(
u
u
=
(
(
(
(
(
(
=
=
= =
=
N
i
i i
N
i
i
N
i
i
N
i
i
N
i
i
b
a
N
(2.4.9)
or
.
1
1
1
1
2
1
1
(
(
(
(
(
u
u
(
(
(
(
(
=
(
=
=
= =
=
N
i
i i
N
i
i
N
i
i
N
i
i
N
i
i
N
b
a
(2.4.10)
The 2 2 matrix inverse is
,
1
1 1 1
1 1 1
2
1
2
1 1
2
(
(
(
(
(
u u
u u
|
|
.
|
\
|
=
(
= = =
= = = =
= =
N
i
i
N
i
i
N
i
i i
N
i
i i
N
i
i
N
i
i
N
i
i
N
i
i
N
i
i
N
N
b
a
(2.4.11)
we define the mean (or the expected values)of and u by writing a line
above and u: thus
,
1
=
=
N
i
i
(2.4.12)
33
.
1
=
u = u
N
i
i
(2.4.13)
The mean is the average value of the data. Now using Eqs. (2.4.12) and
(2.4.13) in Eq. (2.4.11) we obtain
,
2
1
2
1 1
2
2
1 1
2
1 1 1
2
1
N
N
a
N
i
i
N
i
i i
N
i
i
N
i
i
N
i
i
N
i
i i
N
i
i
N
i
i
N
i
i
u u
=
|
|
.
|
\
|
u u
=
=
= =
= =
= = = =
(2.4.14)
.
2
1
2
1
2
1 1
2
1 1 1
N
N
N
N
b
N
i
i
N
i
i i
N
i
i
N
i
i
N
i
i
N
i
i
N
i
i i
u u
=
|
|
.
|
\
|
u u
=
=
=
= =
= = =
(2.4.15)
These can be rewritten in a simpler form by defining the sums of squares
(see Kenney and Keeping [95])
, ) (
2
1
2
1
2
N ss
N
i
i
N
i
i
|
|
.
|
\
|
= =
= =
(2.4.16)
, ) (
2
1
2
1
2
u
|
|
.
|
\
|
u = u u =
= =
uu
N ss
N
i
i
N
i
i
(2.4.17)
, ) )( (
1 1
u
|
|
.
|
\
|
u = u u =
= =
u
N ss
N
i
i i
N
i
i i
(2.4.18)
which are also written as
,
N
ss
V
= (2.4.19)
,
N
ss
V
uu
u
= (2.4.20)
33
. ) , cov(
N
ss
u
= u
(2.4.21)
Here, ) , cov( u is the covariance and
V and
u
V are variances. Note
that the quantities
=
u
N
i
i i
1
and
=
N
i
i
1
2
can also be interpreted as the dot
products
,
1
2
=
=
N
i
i
(2.4.22)
=
u = u
N
i
i i
1
. (2.4.23)
In terms of the sums of squares, the regression coefficient b is given by
.
) , cov(
ss
ss
V
b
u
=
u
= (2.4.24)
and a is given in terms of b using Eq. (2.4.7) as
. b a u = (2.4.25)
The overall quality of the fit is then parameterized in terms of a quantity
known as the correlation coefficient, defined by
,
2
2
uu
u
=
ss ss
ss
cc
(2.4.26)
which gives the proportion of
uu
ss which is accounted for by the
regression.
Let be
i
u
i i
b a + u (2.4.27)
33
then the error between the actual vertical point
i
u and the fitted point is
given by
.
i i i
e u u (2.4.28)
Now define
2
s as an estimator for the variance in
i
e ,
.
2
1
2
2
=
=
N
i
i
N
e
s (2.4.29)
Then s can be given by
.
2 2
2
=
u
uu
u uu
N
ss
ss
ss
N
bss ss
s
(2.4.30)
(see Acton [96]; Gonick and Smith [97]).
The standard errors for a and b are
,
1
) (
2
ss N
s a SE + = (2.4.31)
. ) (
ss
s
b SE = (2.4.32)
Richardson's extrapolation is used to generate high-accuracy results
while using low-order formulas. Although the name attached to the method
refers to a paper written by Richardson and Gaunt in 1927, the idea behind
the technique is much older (see Burden and Faires [98]).
Extrapolation can be applied whenever it is known that an
approximation technique has an error term with a predictable form, one
33
that depends on parameter, usually the step size k . Suppose that for each
number 0 = k we have formula ) (k N that approximates an unknown
value M and that the truncation error involved with the approximation has
the form
, ... ) (
3
3
2
2 1
+ + + = k k k k K K K N M (2.5.1)
for some collection of unknown constants ... , , ,
3 2 1
K K K .
Since the truncation error is ) (k O , we would expect, for example, that
, 01 . 0 ) 01 . 0 ( , 1 . 0 ) 1 . 0 (
1 1
K N M K N M ~ ~ (2.5.2)
and, in general, , ) (
1
k k K N M ~ unless there was a large variation in
magnitude among the constants ... , , ,
3 2 1
K K K .
The object of extrapolation is to find an easy way to combine the rather
inaccurate ) (k O approximations in an appropriate way to produce
formulas with a higher-order truncation error. Suppose, for example, we
can combine the ) (k N formulas to produce an ) (
2
k O approximation
formula, ), (
k N for M with
, ...
) (
3
3
2
2
+ + = k k k K K N M (2.5.3)
for some, again unknown, collection of constants ... ,
2 1
K K . Then we
have
,
0001 . 0 ) 01 . 0 (
01 . 0 ) 1 . 0 (
2 2
K N M K N M ~ ~ (2.5.4)
and so on.
If the constants
1
K and
2
k N
33
approximations in a manner that produces formulas with ) (
3
k O truncation
error, and so on.
To see specifically how we can generate these higher-order formulas,
let us consider the formula for approximating M of the form
. ... ) (
3
3
2
2 1
+ + + + = k k k k K K K N M (2.5.5)
Since the formula is assumed to hold for all positive k , consider the result
when we replace the parameter k by half its value. Then we have the
formula
. ...
8 4 2 2
3
3
2
2 1
+ + + +
|
.
|
\
|
=
k k k k
K K K N M (2.5.6)
Subtracting Eq. (2.5.3) form twice this equation eliminates the term
involving
1
K and gives
. ...
4 2
) (
2
2
3
3
3
2
2
2
+
|
|
.
|
\
|
+
|
|
.
|
\
|
+
(
|
.
|
\
|
= k
k
k
k
k
k
K K N N M (2.5.7)
To facilitate the discussion, we define ) ( ) (
1
k k N N and
. ) (
2 2
) (
2
2 ) (
1 1 1 1 1 2
(
|
.
|
\
|
+
|
.
|
\
|
=
(
|
.
|
\
|
= k
k k
k
k
k N N N N N N (2.5.8)
Then we have the ) (
2
k O approximation formula for M:
. ...
4
3
2
) (
3
3
2
2
2
= k k k
K K
N M (2.5.9)
If we now replace k by
2
k
in this formula, we have
. ...
32
3
8 2
3
3
2
2
2
|
.
|
\
|
= k k
k K K
N M (2.5.10)
30
This can be combined with Eq. (2.5.9) to eliminate the
2
k term.
Specifically, subtracting Eq. (2.5.9) from 4 times Eq. (2.5.10) gives
, ...
8
3
) (
2
4 3
3
3
2 2
+ +
|
.
|
\
|
= k k
k K
N N M (2.5.11)
and dividing by 3 gives an ) (
3
k O formula for approximating M:
. ...
8 3
) (
2
2
3 3
2 2
2
+ +
(
(
(
(
|
.
|
\
|
+
|
.
|
\
|
= k
k
k
k K
N N
N M (2.5.12)
By defining
.
3
) (
2
2
) (
2 2
2 3
k
k
k
k
N N
N N
|
.
|
\
|
+
|
.
|
\
|
= (2.5.13)
We have the ) (
3
k O formula
. ...
8
) (
3 3
3
+ + = k k
K
N M . (2.5.14)
The process is continued by constructing an ) (
4
k O approximation
,
7
) (
2
2
) (
3 3
3 4
k
k
k
k
N N
N N
|
.
|
\
|
+
|
.
|
\
|
= (2.5.15)
and an ) (
5
k O approximation
,
15
) (
2
2
) (
4 4
4 5
k
k
k
k
N N
N N
|
.
|
\
|
+
|
.
|
\
|
= (2.5.16)
and so on. In general, if M can be written in the form
33
), ( ) (
1
1
m
m
j
j
j
O K N M k k k + + =
=
(2.5.17)
then for each m j ..., , 3 , 2 = , we have an ) (
j
O k approximation of the form
.
1 2
) (
2
2
) (
1
1 1
1
|
.
|
\
|
+
|
.
|
\
|
=
j
j j
j j
N N
N N
k
k
k
k (2.5.18)
55
* The results of this chapter is already appeared in Ref. [99].
In this chapter, a unified governing equation is firstly derived
from the basic equations of the rotating variable-thickness solid
disk and the proposed stress-strain relationship. The outer edge of
the solid disk is considered to have free boundary conditions. The
analytical solution for rotating solid disk with arbitrary cross-
section of continuously variable-thickness is presented. Next, the
finite difference method (FDM) is introduced to solve the
governing equation. A comparison between both analytical and
numerical solutions is made. Finally, some application examples
are given to demonstrate the validity of the proposed method.
As the effect of thickness variation of rotating solid disks can be taken
into account in their equation of motion, the theory of the variable-
thickness solid disks can give good results as that of uniform-thickness
disks as long as they meet the assumption of plane stress. The present solid
disk is considered as a single layer of variable thickness. After considering
this effect, the equation of motion of rotating disks with variable thickness
can be written as
( ) , 0 ) ( ) ( ) (
d
d
2 2
= + r r h r h r r h
r
r
e o o
u
(3.1.1)
65
where
r
o and
u
o are the radial and circumferential stresses, h(r) is the
variable-thickness of the disk, r is the radial coordinate, is the material
density of the rotating solid disk and e is the constant angular velocity.
The relations between the radial displacement u and the strain
components are irrespective of the thickness of the rotating solid disk.
They can be written as
, ,
d
d
r
u
r
u
r
= =
u
c c (3.1.2)
where
r
c and
u
c are the radial and circumferential strains, respectively.
The above geometric relations lead to the following condition of
deformation harmony:
. 0 ) (
d
d
=
r
r
r
c c
u
(3.1.3)
For the elastic deformation, the constitutive equations for the variable-
thickness solid disk can be described with Hooke's law
, ,
E E
r r
r
vo o
c
vo o
c
u
u
u
=
= (3.1.4)
where E is Young's modulus and v is Poisson's ratio. Introducing the stress
function and assuming that the following relations hold between the
stresses and the stress function
.
d
d
) (
1
,
) (
2 2
r
r r h r r h
r
e
o
u
+ = = (3.1.5)
Substituting Eq. (3.1.5) into Eq. (3.1.4), one obtains
65
.
) ( d
d
) (
1 1
,
d
d
) ( ) (
1 1
2 2
2 2
(
+
|
|
.
|
\
|
=
(
|
|
.
|
\
|
=
r
r r h r r h E
r
r r h r r h E
r
e
v
c
ve
v
c
u
(3.1.6)
The substitution of Eq. (3.1.6) into Eq. (3.1.3) produces the following
confluent hypergeometric differential equation for the stress function
: ) (r
|
.
|
\
|
+
+
|
.
|
\
|
|
|
|
|
.
|
\
|
) (
d
d
) (
) 1 (
) (
d
d
) (
) (
) (
d
d
2
r
r r Eh
r
r r r h
r h
r h
r
E
r
v
) (
) (
) 1 (
) (
) ( ) (
) (
d
d
2 2
r
r r Eh
r
r r h r r h
r h
r
E
r
v
v
+
|
|
|
|
.
|
\
|
+
|
.
|
\
|
+
. 0 2
) (
) (
d
d
2 2
2
2
2
2 2
= +
|
|
|
|
|
.
|
\
|
+ + +
E
r
r
r h
r
r
E
r
E
r ve
e
e
(3.2.1)
The above equation may be simplified to be
2
2
2
2
) (
) (
d
d
) (
d
d
) (
) (
) ( 2 ) (
d
d
r Eh
r
r
r h
r
r r h
r Eh
r rh r
r
r
|
.
|
\
|
|
.
|
\
|
|
.
|
\
|
+
|
|
.
|
\
|
+
e
. 0
) 1 (
) (
) ( ) ( ) (
d
d
2 2
2
=
+
+
|
.
|
\
|
+
|
.
|
\
|
E
r
r rEh
r r h r r h
r e v
v
(3.2.2)
65
Then, multiplying by ) (r rEh , one obtains
. 0 ) 3 (
d
d
1
d
d
d
d
1
d
d
3 2
2
2
2
= + +
|
.
|
\
|
|
.
|
\
|
+ r h
r
h
h
r
r r
h
h
r
r
r
r e v
v
(3.2.3)
The boundary conditions for the rotating solid disk are
. at 0
, 0 at
b r
r
r
r
= =
= =
o
o o
u
(3.2.4)
The thickness of the solid disk is assumed to vary nonlinearly through
the radial direction. It is assumed to be in terms of a simple exponential
power law distribution according to the following case:
, e ) (
0
k
b
r
n
h r h
|
.
|
\
|
= (3.2.5)
Figure 3.1 (a): Variable-thickness solid disk profiles for k = 0.7 and n = 2.
65
Figure 3.1 (b): Variable-thickness solid disk profiles for k = 1.5 and n = 2.
Figure 3.1 (c): Variable-thickness solid disk profiles for k = 2.5 and n = 0.5.
56
where
0
h is the thickness at the middle of the disk, n and k are geometric
parameters and b is the outer radius of the disk (see Figure 3.1). The value
of n equal to zero represents a uniform-thickness solid disk while the value
of k equal to unity represents a linearly decreasing variable-thickness solid
disk. For small k and large n (k = 0.7 or 1.5 and n = 2) the profile of the
solid disk is concave while it is convex for large k and small n (k = 2.5 and
n = 0.5). It is to be noted that the parameter n determines the thickness at
the outer edge of the solid disk relative to
0
h while the parameter k
determine the shape of the profile.
Introducing the following dimensionless forms:
b r R / = ,
, ) 3 ( v e + = O b
), (
1
) (
2
0
r
bh
R
O
= u
), , ( ) , (
2
2 1 u
c c c c
r
E
O
=
). , (
1
) , (
2
2 1 u
o o o o
r
O
= (3.2.6)
Then, Eq. (3.2.3) may be written in the following simple form
. 0 e ) 1 (
d
d
) 1 (
d
d
3
2
2
2
= + u +
u
+ +
u
R R kn
R
R knR
R
R
k
nR k k
v (3.2.7)
The general solution of the above equation can be written as
= u
}
R
j i j i
R
W F C R M R R
k n k
d ) ( ) ( ) ( e ) (
, 1 ,
2 2
, d ) ( ) ( ) (
, 2 ,
)
`
+ +
}
R
j i j i
M F C R W (3.2.8)
56
where
1
C and
2
C are arbitrary constants, is a dummy parameter,
j i
M
,
and
j i
W
,
are Whittaker's functions
), , , ( ) ( ), , , ( ) (
, ,
k
j i
k
j i
nR j i W R W nR j i M R M = = (3.2.9)
in which
. 0 ,
1
,
2
1
> = = k
k
j
k
i
v
(3.2.10)
In addition, the function ) (R F is given in terms of Whittaker's functions
by
.
) ( ) ( ) ( ) ( ) 1 (
e
) (
, 1 , , 1 ,
2
2 2
R W R kM R M R W
R
R F
j i j i j i j i
R
k n k
+ +
+
+
=
v
(3.2.11)
The substitution of Eq. (3.2.8) into Eq. (3.1.5) with the aid of the
dimensionless forms given in Eq. (3.2.6) gives the radial and
circumferential stresses in the following forms:
=
}
R
j i j i
R
W F C R M R R
k n k
o d ) ( ) ( ) ( e ) (
, 1 ,
1
1
2 2
, d ) ( ) ( ) (
, 2 ,
)
`
+ +
}
R
j i j i
M F C R W (3.2.12)
.
3 d
d
e ) (
2
2
v
o
+
+
u
=
R
R
R
k
nR
(3.2.13)
Here, the first derivative of the stress function ) (R u with respect to R may
be given easily by using Eq. (3.2.8). Note that the first derivatives of
Whittaker's functions
j i
M
,
and
j i
W
,
can be represented by
56
( ) ( ) | |
( ) | |. ) ( ) ( ) (
d
d
, ) ( ) ( ) (
d
d
, 1 ,
2
1
,
, 1
2
1
,
2
1
,
R W R W i nR
R
k
R W
R
R M j i R M i nR
R
k
R M
R
j i j i
k
j i
j i j i
k
j i
+
+
=
+ + + =
(3.2.14)
Finally, the dimensionless strains and the corresponding radial
displacement may be obtained easily using Eq. (3.2.12) and Eq. (3.2.13) as
well as the dimensionless forms given in Eq. (3.2.6). Therefore, all of
stress function, stresses, strains, and radial displacement may be
determined completely after applied the dimensionless of the boundary
conditions given in Eq. (3.2.4).
The finite difference method presented in Chapter 2 is used here to
solve the present problem numerically. The resolution of the elastic
problem of rotating solid disk with variable thickness is to solve a second-
order differential equation (3.2.7) under the given boundary conditions
0 ) 1 ( ) 0 ( = u = u such that ) 0 ( ) 0 (
2 1
o o = . Equation (3.2.7) can be written
in the following general form:
), ( ) ( ) ( R s R q R p + u + u' = u' ' (3.3.1)
where the prime ) (' denotes differentiation with respect to R and
. e ) ( ,
1
) ( ,
1
) (
2
R R s
R
R kn
R q
R
knR
R p
k
nR
k k
=
+
=
+
=
v
(3.3.2)
It is clear that the above problem has a unique solution because
), ( ), ( R q R p and ) (R s are continuous on [0, 1] and 0 ) ( > R q on [0, 1]. The
linear second-order boundary value problem given in Eq. (3.3.1) requires
that difference-quotient approximations be used for approximating u' and
u' ' . First we select an integer 0 > N and divided the interval [0, 1] into
56
) 1 ( + N equal subintervals, whose end points are the mesh points
, R i R
i
A = for , 1 , ... , 1 , 0 + = N i where ) 1 /( 1 + = A N R . At the interior mesh
points, ,
i
R , , ... , 2 , 1 N i = the differential equation to the approximated is
). ( ) ( ) ( ) ( ) ( ) (
i i i i i i
R s R R q R R p R + u + u' = u' ' (3.3.3)
If we apply the centered difference approximations of ) (
i
R u' and ) (
i
R u' '
to Eq. (3.3.3), we arrive at the system:
| |
1
2
1
) (
2
1 ) ( ) ( 2 ) (
2
1
+
u
(
A
u A + + u
(
A
+
i i i i i i
R p
R
R q R R p
R
), ( ) (
2
i
R s R A = (3.3.4)
for each . ..., , 2 , 1 N i = The N equations, together with the boundary
conditions
, 0
, 0
1
0
= u
= u
+ N
(3.3.5)
are sufficient to determine the unknowns ,
i
u 1 ..., , 2 , 1 , 0 + = N i . The
resulting system of Eq. (3.3.4) is expresses in the tri-diagonal N N -
matrix form:
, B A = u (3.3.6)
where
. ..., , 2 , 1 ), ( ) (
, 2 , ..., , 4 , 3 , , 2 ..., , 2 , 1 , 0
, ..., , 3 , 2 ), (
2
1
, 1 ..., , 2 , 1 ), (
2
1
, ..., , 2 , 1 ), ( ) ( 2
2
, ,
1 ,
1 ,
2
,
N i R s R B
i j N j N i A A
N i R p
R
A
N i R p
R
A
N i R q R A
i i
i j j i
i i i
i i i
i i i
= A =
+ > = = = =
=
A
+ =
=
A
=
= A =
+
(3.3.7)
56
The solution of the finite difference discretization of the two-point linear
boundary value problem can therefore be found easily even for very small
mesh sizes.
Some numerical examples for the rotating variable-thickness solid disks
will be given according the analytical and numerical solutions ) 3 . 0 ( = v .
According to Eq. (3.2.6), the stress function, radial displacement, strains
and stresses of the rotating variable-thickness solid disk are determined as
per the analytical solution are compared with those obtained by the
numerical FDM solution.
The results of the present FDM investigations for the stress function u
for the rotating variable-thickness solid disk with k = 2.5 and n = 0.5 are
reported in Table 3.1. For this example, N = 9, 19, 39 and 79, so R A has
the corresponding values 0.1, 0.05, 0.025 and 0.0125, respectively. The
FDM gives results compared well with the exact solution, especially for
small values of R A . The relative error between the exact method and the
FDM with 0125 . 0 = AR at 6 . 0 =
i
R for example, may be less than
5
10 62 . 2
.
56
Table 3.1: Dimensionless stress function u of a rotating variable-
thickness solid disk (k = 2.5, n = 0.5).
Analytical
FDM
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0 0 0 0 0 0
0.001318247
0.002635012
0.003948803
0.005258119
0.006561453
0.007857295
0.009144132
0.010420449
---
---
---
---
---
---
---
0.010333909
---
---
---
0.005341761
---
---
---
0.010398523
---
0.002632275
---
0.005253982
---
0.007852303
---
0.010414946
0.001317819
0.002634312
0.003947918
0.005270818
0.006560297
0.007856045
0.009142869
0.010419072
0.0125
0.0250
0.0375
0.0500
0.0625
0.0750
0.0875
0.1000
0.011684731
0.012935466
0.014171142
0.015390252
0.016591293
0.017772771
0.018933196
0.020071091
---
---
---
---
---
---
---
0.019984044
---
---
---
0.015367662
---
---
---
0.020049282
---
0.012929711
---
0.015384444
---
0.017767073
---
0.020065635
0.011683316
0.012934026
0.014169696
0.015388799
0.016589850
0.017771346
0.018931799
0.020069727
0.1125
0.1250
0.1375
0.1500
0.1625
0.1750
0.1875
0.2000
0.021184986
0.022273424
0.023334964
0.024368176
0.025371649
0.026343989
0.027283820
0.028189791
---
---
---
---
---
---
---
0.028132394
---
---
---
0.024349525
---
---
---
0.028175489
---
0.022268320
---
0.024363514
---
0.026339842
---
0.028186218
0.021183663
0.022272149
0.023333741
0.024367011
0.025370546
0.026342953
0.027282855
0.028188899
0.2125
0.2250
0.2375
0.2500
0.2625
0.2750
0.2875
0.3000
0.029060568
0.029894846
0.030691343
0.031448803
0.032166002
0.032841742
0.033474861
0.034064227
---
---
---
---
---
---
---
0.034048816
---
---
---
0.031439588
---
---
---
0.034060453
---
0.029891894
---
0.031446504
---
0.032840118
---
0.034063288
0.029059753
0.029894109
0.030690687
0.031448230
0.032165512
0.032841338
0.033474542
0.034063994
0.3125
0.3250
0.3375
0.3500
0.3625
0.3750
0.3875
0.4000
0.034608743
0.035107349
0.035559023
0.035962780
0.036317677
0.036622811
0.036877324
0.037080400
---
---
---
---
---
---
---
0.037107585
---
---
---
0.035964469
---
---
---
0.037087271
---
0.035107098
---
0.035963207
---
0.036623899
---
0.037082122
0.034608596
0.035107288
0.035559047
0.035962888
0.036317868
0.036623085
0.036877677
0.037080832
0.4125
0.4250
0.4375
0.4500
0.4625
0.4750
0.4875
0.5000
55
Table 3.1: Continued
Analytical
FDM
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0.037231269
0.037329209
0.037373544
0.037363648
0.037298942
0.037178902
0.037003051
0.036770967
0.037231777
0.037329791
0.037374198
0.037364369
0.037299729
0.037179570
0.037003958
0.036771929
---
0.037331532
---
0.037366529
---
0.037182291
---
0.036774810
---
---
---
0.037375156
---
---
---
0.036786324
---
---
---
---
---
---
---
0.036832186
0.5125
0.5250
0.5375
0.5500
0.5625
0.5750
0.5875
0.6000
0.036482282
0.036136679
0.035733896
0.035273729
0.034756025
0.034180690
0.033547685
0.032857027
0.036483294
0.036137738
0.035734998
0.035274868
0.034757198
0.034181892
0.033548911
0.032858271
---
0.036140911
---
0.035278284
---
0.034185494
---
0.032862001
---
---
---
0.035291940
---
---
---
0.032876919
---
---
---
---
---
---
---
0.032936503
0.6125
0.6250
0.6375
0.6500
0.6625
0.6750
0.6875
0.7000
0.032108789
0.031303102
0.030440152
0.029520183
0.028543493
0.027510437
0.026421427
0.025276927
0.032110047
0.031304368
0.030441421
0.029521449
0.028544751
0.027511682
0.026422653
0.025278128
---
0.031308165
---
0.029525248
---
0.027515416
---
0.025281728
---
---
---
0.029540441
---
---
---
0.025296129
---
---
---
---
---
---
---
0.025353741
0.7125
0.7250
0.7375
0.7500
0.7625
0.7750
0.7875
0.8000
0.024077458
0.022823594
0.021515961
0.020155240
0.018742161
0.017277507
0.015762106
0.014196841
0.024078628
0.022824727
0.021517051
0.020156281
0.018743148
0.017278432
0.015762966
0.014197627
---
0.022828123
---
0.020159404
---
0.017281209
---
0.014199988
---
---
---
0.020171897
---
---
---
0.014209433
---
---
---
---
---
---
---
0.014247568
0.8125
0.8250
0.8375
0.8500
0.8625
0.8750
0.8875
0.9000
0.012582635
0.010920462
0.009211340
0.007456327
0.005656526
0.003813078
0.001927164
0
0.012583344
0.010921087
0.009211874
0.007456766
0.005656864
0.003813309
0.001927282
0
---
0.010922960
---
0.007458083
---
0.003814000
---
0
---
---
---
0.007463350
---
---
---
0
---
---
---
---
---
---
---
0
0.9125
0.9250
0.9375
0.9500
0.9625
0.9750
0.9875
1.0000
55
Richardson extrapolation method is applied here with , 05 . 0 , 1 . 0 = AR
025 . 0 and 0.0125 the obtained results are listed in Table 3.2. These
extrapolations are given, respectively by
,
3
) 1 . 0 ( ) 05 . 0 ( 4
Ext
1
= A u = A u
=
R R
i i
i
(3.4.1a)
,
3
) 05 . 0 ( ) 025 . 0 ( 4
Ext
2
= A u = A u
=
R R
i i
i
(3.4.1b)
,
3
) 025 . 0 ( ) 0125 . 0 ( 4
Ext
3
= A u = A u
=
R R
i i
i
(3.4.1c)
.
15
Ext Ext 16
Ext
1 2
4
i i
i
= (3.4.1d)
Table 3.2 shows that all extrapolations results are correct to the decimal
places listed. In fact, if sufficient digits are maintained, the approximation
of
i 4
Ext gives results those agree with the exact solution with maximum
difference error of
9
10 0 . 1
at some of the mesh points.
Table 3.2: Dimensionless stress function u of a rotating variable-
thickness solid disk using Richardson's extrapolation method with
different values of R (k = 2.5, n = 0.5).
i
R
i 1
Ext
i 2
Ext
i 3
Ext
i 4
Ext
Analytical
0.0 0 0 0 0 0
0.1 0.010420061 0.010420421 0.010420447 0.010420444 0.010420449
0.2 0.020071028 0.020071086 0.020071091 0.020071090 0.020071091
0.3 0.028189855 0.028189794 0.028189792 0.028189790 0.028189791
0.4 0.034064332 0.034064233 0.034064229 0.034064227 0.034064227
0.5 0.037080500 0.037080406 0.037080401 0.037080400 0.037080400
0.6 0.036771037 0.036770972 0.036770969 0.036770967 0.036770967
0.7 0.032857058 0.032857029 0.032857028 0.032857028 0.032857027
0.8 0.025276926 0.025276927 0.025276928 0.025276927 0.025276927
0.9 0.014196825 0.014196839 0.014196840 0.014196840 0.014196841
1.0 0 0 0 0 0
55
Table 3.3: Dimensionless stress function u of a rotating variable-
thickness solid disk (k = 0.7, n = 2).
Analytical
FDM
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0 0 0 0 0 0.0
0.003891568 0.003891627 0.003891774 0.003892229 0.003893631 0.1
0.006897877 0.006898133 0.006898890 0.006901872 0.006913732 0.2
0.008971625 0.008972022 0.008973209 0.008977938 0.008996897 0.3
0.010102762 0.010103227 0.010104619 0.010110178 0.010132490 0.4
0.010321685 0.010322155 0.010323563 0.010329193 0.010351793 0.5
0.009683519 0.009683947 0.009685228 0.009690352 0.009710919 0.6
0.008256930 0.008257280 0.008258329 0.008262527 0.008279371 0.7
0.006116796 0.006117044 0.006117787 0.006120761 0.006132695 0.8
0.003339509 0.003339638 0.003340026 0.003341577 0.003347797 0.9
0 0 0 0 0 1.0
Table 3.4: Dimensionless stress function u of a rotating variable-
thickness solid disk (k = 1.5, n = 2).
Analytical
FDM
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0 0 0 0 0 0.0
0.005416705 0.005416605 0.005416315 0.005415269 0.005412300 0.1
0.009932747 0.009933554 0.009935982 0.009945759 0.009985706 0.2
0.013033300 0.013034871 0.013039590 0.013058534 0.013135277 0.3
0.014512487 0.014514484 0.014520482 0.014544543 0.014641868 0.4
0.014415043 0.014417117 0.014423346 0.014448330 0.014549332 0.5
0.012959495 0.012961364 0.012966973 0.012989467 0.013080378 0.6
0.010460010 0.010461482 0.010465900 0.010483617 0.010555201 0.7
0.007259771 0.007260750 0.007263688 0.007275470 0.007323059 0.8
0.003682061 0.003682530 0.003683938 0.003689581 0.003712369 0.9
0 0 0 0 0 1.0
55
Tables 3.3 and 3.4 present the results of investigations for the stress
function u for the rotating variable-thickness solid disks with k = 0.7, n =
2 and k = 1.5, n = 2, respectively.
Now the least square method and curve fitting are used for the discrete
results of the stress function u. So, one can get easily the radial
displacement, strains and stresses since we have u as a continuous
function of R. The distributions of the stress function, radial and
circumferential stresses are presented in Figure 3.2. The numerical FDM
solution is compared with the exact analytical solution for the rotating
variable-thickness solid disk with k = 2.5 and n = 0.5. It can be seen that
the FDM can describe the stress function, radial displacement, strains and
stresses through the thickness of the rotating solid disk very well enough.
For the sake of completeness and accuracy, additional results for the
stress function, radial stress and circumferential stress are presented in
Figures 3.3-3.5 for different values of the geometric parameters k and n.
Figure 3.3 shows the stress function u through the radial direction of the
rotating solid disk with k = 2.5, n = 0.5; k = 0.7, n = 2 and k = 1.5, n = 2.
Similar results for the radial
1
o and the circumferential
2
o stresses are
plotted in Figures 3.4 and 3.5. Figure 3.3 shows that the stress function u
increases as k increases and this irrespective of the value of n. Figures 3.4
and 3.5 show that k = 2.5, n = 0.5 gives the largest stresses. The
intersection of the two cases k = 0.7, n = 2 and k = 1.5, n = 2 may be
occurred at R = 0.1 for the radial stress and at R = 0.15 for the
circumferential stress.
56
Figure 3.2: Stress function u, radial stress
1
o and circumferential stress
2
o for the variable-thickness solid disk.
Figure 3.3: Stress function u of the variable-thickness solid disk for
different values of k and n.
56
Figure 3.4: Radial stress
1
o in the variable-thickness solid disk for
different values of k and n.
Figure 3.5: Circumferential stress
2
o in the variable-thickness solid disk
for different values of k and n.
56
Figure 3.6: Radial displacement U of the variable-thickness solid disk for
different values of k and n.
Figure 3.7: Radial strain
1
c in the variable-thickness solid disk for
different values of k and n.
56
Figure 3.8: Circumferential strain
2
c in the variable-thickness solid disk
for different values of k and n.
The radial displacement, radial strain as well as circumferential strain are
also presented in Figures 3.6-3.8 for different values of the geometric
parameters k and n. Figure 3.6 shows the radial displacement U through the
radial direction of the rotating solid disk for different values of k and n.
Similar results for the radial strain
1
c and the circumferential strain
2
c are
plotted in Figures 3.7 and 3.8. We notice as show in Figure 3.6 that, the
radial displacement U increases as k increases and this irrespective of the
value of n. Figures 3.7 and 3.8 show that k = 2.5, n = 0.5 gives the largest
strains. The intersection of the two cases k = 0.7, n = 2 and k = 1.5, n = 2
may be occurred at R = 0.08 for the radial strain and at R = 0.18 for the
circumferential strain.
56
It is clear that, the FDM gives stress function, radial strain and stresses
consequently, with excellent accuracy with the exact analytical solution. In
fact, FDM may be failed to get accurate radial displacement and
circumferential strain, especially, at the outer edges of the solid disk (see
Figures 3.6 and 3.8). However, in some cases of rotating variable-thickness
solid disks, the analytical solutions are not available. So, one can trustily
use the present FDM solutions.
The rotating solid disk with variable thickness is treated herein. By
introducing a suitable stress function, the governing equation is derived
from the equation of motion of rotating disk, compatibility equation and
the proposed stress-strain relationship. Both the analytical and numerical
solutions are presented. The calculation of the rotating solid disk is turned
into finding the solution of a second-order differential equation under the
given conditions at the center and the outer edge of the disk. The numerical
solution is based upon the finite difference method. The governing
equation is solved analytically with the help of Whittaker's functions and a
number of numerical examples are studied. The results of the two solutions
at different disk configurations are compared. The proposed FDM
approach gives very agreeable results to the analytical solution and so it
may be used for different problems that analytical solutions are not
available.
75
* The results of this chapter is already sent for possible publication in a suitable journal.
In this chapter, two different annular disks for the radially varying
thickness are given. The numerical finite difference method
(FDM) and modified Runge-Kutta's method (R-K) solutions as
well as the analytical solution are available for the first disk while
the analytical solution is not available for the second annular disk.
Both analytical and numerical results for radial displacement,
stresses and strains are obtained for the first annular disk of
variable thickness. The accuracy of the present numerical
solutions is discussed and their ability of use for the second
rotating variable-thickness annular disk is investigated. Finally,
the distributions of radial displacement, stresses and strains are
presented and the appropriate comparisons and discussions are
made at the same angular velocity.
As the effect of thickness variation of rotating annular disks can be
taken into account in their equation of motion, the theory of the variable-
thickness annular disks can give good results as that of the uniform-
thickness annular disks as long as they meet the assumption of plane stress.
67
After considering this effect, the equation of motion of rotating disks with
variable thickness can be written as
( ) , 0 ) ( ) ( ) (
d
d
2 2
= + r r h r h r r h
r
r
e o o
u
(4.1.1)
where
r
o and
u
o are the radial and circumferential stresses, r is the radial
coordinate, is the density of the rotating disk, e is the constant angular
velocity, and h(r) is the thickness which is function of the radial coordinate
r.
The relations between the radial displacement u and the strains are
irrespective of the thickness of the rotating disk. They can be written as
, ,
d
d
r
u
r
u
r
= =
u
c c (4.1.2)
where
r
c and
u
c are the radial and circumferential strains, respectively.
For the elastic deformation, the constitutive equations for the rotating
disk can be described with Hooke's law
, ,
E E
r r
r
vo o
c
vo o
c
u
u
u
=
= (4.1.3)
where E is Young's modulus and v is Poisson's ratio.
The substitution of Eq. (4.1.2) into Eq. (4.1.3) produces the constitutive
equations for
r
o and
u
o as:
.
1
,
1
2
2
|
|
.
|
\
|
+
=
|
|
.
|
\
|
+
=
dr
du
r
u
E
r
u
dr
du E
r
v
v
o
v
v
o
u
(4.1.4)
The thickness of the annular disk is given using various distributions
through the radial direction as follows:
66
Figure 4.1 (a): Variable-thickness annular disk 1 profile for k = 0.4.
Figure 4.1 (b): Variable-thickness annular disk 1 profile for k = 4 . 0 .
67
Figure 4.1 (c): Variable-thickness annular disk 1 profile for k = 0.2.
Figure 4.1 (d): Variable-thickness annular disk 1 profile for k = 2 . 0 .
68
Disk 1: Nonlinearly distribution
. ) (
2
0
k
b
r
h r h
|
.
|
\
|
= (4.1.5)
Disk 2: Exponentially distribution
. e 2 ) (
0
|
|
|
|
|
.
|
\
|
=
(
(
|
.
|
\
|
|
.
|
\
|
k k
b
a
b
r
n
h r h (4.1.6)
For both disks,
0
h is the thickness at the inner edge of the disk, n and k are
geometric parameters, a is the inner radius of the disk and b is the outer
radius of the disk. The value of n equal to zero represents a uniform-
thickness annular disk. It is to be noted that the parameter n determines the
thickness at the outer edge of the annular disk relative to
0
h while the
parameter k determine the shape of the profile. For disk 1, the geometric
parameter k is given according to (k = 2 . 0 ) and (k = 4 . 0 ) For positive
values of k, the profile is concave whereas it may be convex for negative
values of k (see Figure 4.1). For disk 2, the geometric parameters k and n
are given according to three different sets. For small k and large n (k = 1.4,
n = 2, and k = 0.8, n = 1.2) the profile of the annular disk is convex while it
is concave for large k and small n (k = 2, n = 0.6) (see Figure 4.2).
To simplify the solving process, we introduce the following
dimensionless variables:
). , (
1
) , ( , / , /
), , ( ) , ( ), ( , ) 1 (
2
2
2 1
2
2 1
2
2
u
u
o o
v
o o
c c c c v e
r
r
b a A b r R
E
r u
b
E
U b
O
= = =
O
=
O
= = O
(4.1.7)
78
Figure 4.2 (a): Variable-thickness annular disk 2 profile for k = 1.4
and n = 2.
Figure 4.2 (b): Variable-thickness annular disk 2 profile for k = 0.8
and n = 1.2.
78
Figure 4.2 (c): Variable-thickness annular disk 2 profile for k = 2
and n = 0.6.
The substitution of Eq. (4.1.5) into Eq. (4.1.1) with the aid of
r
o and
u
o given in Eq. (4.1.4) produces the following confluent hypergeometric
differential equation for the radial displacement ) (r u according to the
annular disk 1
( )
|
.
|
\
|
|
.
|
\
|
+
|
|
.
|
\
|
|
.
|
\
|
) (
1
2 1 ) (
1
2
2
0
2
2
2
2
0
r u
dr
d
E
b
r
h
k r u
dr
d
rE
b
r
h
k k
v v
( ) . 0 ) (
) 1 (
1 2
2 2
2
0
2
2
0
=
|
.
|
\
|
+
|
.
|
\
|
+
r
b
r
h r u
r
E
b
r
h
k
k
k
e
v
v (4.1.8)
78
Then multiplying (4.1.8) by
|
|
.
|
\
|
|
.
|
\
|
E
b
r
h
r
k 2
0
2
) 1 ( v
, we obtain
) ( ) 1 2 ( ) ( ) 1 2 ( ) (
2
2
2
r u k r u
dr
d
r k r u
dr
d
r +
|
.
|
\
|
|
|
.
|
\
|
v
. 0
) 1 (
2 2 3
=
+
E
r e v
(4.1.9)
Using the dimensionless variables given in (4.1.7) we obtain
E
U k b
E
U
dR
d
R k b
E
U
dR
d
b R
) 1 2 (
) 1 2 (
2
2
2
2
2 2
+ O
|
.
|
\
|
O
+
|
|
.
|
\
|
O
v
. 0
2 3
=
O
+
E
b R
(4.1.10)
Multiplying (4.1.10) by
2
O b
E
gives
. 0 ) 1 2 (
d
d
) 1 2 (
d
d
3
2
2
2
= + + + R U k
R
U
R k
R
U
R v (4.1.11)
In a similar way, we can obtain the following differential equation for the
annular disk 2. By the substitution of Eq. (4.1.6) into Eq. (4.1.1) with the
aid of
r
o and
u
o given in Eq. (4.1.4), we get
( )
|
.
|
\
|
|
|
.
|
\
|
+
|
.
|
\
|
) (
1
) ( 2 ) (
2
0 0
r u
dr
d E
r f h r f k
b
r
n h
k
v
( ) ) (
) 1 (
) ( 2 ) (
2
0 0
r u
r
E
r f h r f k
b
r
n h
k
v
v
|
|
.
|
\
|
|
.
|
\
|
+
78
( )
( )
, 0
1
) ( ) ( 2
) ( 2
2
2
2
0
2 2
0
=
|
|
.
|
\
|
+ +
v
e
r u
dr
d
rE r f h
r r f h (4.1.12)
where
. ) (
|
|
.
|
\
|
|
.
|
\
|
|
.
|
\
|
=
k k
b
a
b
r
n
e r f (4.1.13)
Then multiplying (4.1.12) by
( )E r f h
r
) ( 2
) 1 (
0
2
v
we obtain the following simple form:
|
.
|
\
|
|
|
.
|
\
|
+
|
.
|
\
|
|
|
.
|
\
|
) (
2 ) (
) ( 2 ) (
) (
2
2
2
r u
dr
d
r
r f
r f r f
b
r
nk
r u
dr
d
r
k
. 0
) 1 (
) (
2 ) (
) ( 2 ) (
2 2 3
=
|
|
.
|
\
|
+
|
.
|
\
|
E
r
r u
r f
r f r f
b
r
nk
k
e v
v
(4.1.14)
Using the dimensionless variables given in (4.1.7) we obtain
U
e E
e ke nR b
dR
dU
e
ke nR
e E
R b
dR
U d
E
k k
k k k k
k k
k k
k k
A R n
A R n A R n k
A R n
A R n k
A R n
b R
|
.
|
\
|
+
|
.
|
\
|
+ O
|
.
|
\
|
|
.
|
\
|
|
.
|
\
|
+
O
|
|
.
|
\
|
+
+ +
+
+
+
O
) (
) ( ) ( 2
) (
) (
) (
2
2
2
2
2
2
2
2 2
v
. 0
2 3
=
O
+
E
R
(4.1.15)
78
Multiplying (4.1.15) by
2
O b
E
gives the final differential equation for disk 2
as follows
R
U
R
e e
e e e knR
R
U
R
k k
k k k
nA nR
nA nR nA k
d
d
2
2
d
d
2
2
2
|
|
|
.
|
\
|
+
+
. 0
2
2
3
= +
|
|
|
.
|
\
|
+
+ R U
e e
e e e R kn
k k
k k k
nA nR
nA nR nA k
v
(4.1.16)
The analytical solution for Eq. (4.1.11) may be available while it is not
available for Eq. (4.1.16). The modified numerical solutions may be
available for both cases. Firstly, we will get both the analytical and
numerical solutions for Eq. (4.1.11). If the numerical solutions are
compared will with the analytical, we can use them to solve Eq. (4.1.16)
for the second case.
The analytical solution for Eq. (4.1.11) concerning disk 1 can be written as
), ( ) ( ) ( ) (
2 2 1 1
R P R F c R F c R U + + = (4.2.1)
where
1
c and
2
c are arbitrary constants and ) (
1
R F and ) (
2
R F are given
by:
, ) (
2 2 2 1
1
2
v k k k k
R R F
+ + +
=
( ) ,
2 2 2 1
2
2
v k k k k
R R F
+ + + +
=
( )
.
2 6 2
) (
3
=
k
R
R P
v
(4.2.2)
78
The substitution of Eq. (4.2.2) into Eq. (4.1.4) with the aid of the
dimensionless forms given in Eq. (4.1.7) gives the radial and
circumferential stresses in the following forms:
.
) ( ) ( ) ( ) ( ) ( ) (
) (
,
) ( ) ( ) ( ) ( ) ( ) (
) (
2 2
2
1 1
1 2
2 2
2
1 1
1 1
R
R P
dR
R dP
R
R F
dR
R dF
c
R
R F
dR
R dF
c R
R
R P
dR
R dP
R
R F
dR
R dF
c
R
R F
dR
R dF
c R
+ +
|
.
|
\
|
+ +
|
.
|
\
|
+ =
+ +
|
.
|
\
|
+ +
|
.
|
\
|
+ =
v v v o
v v v o
(4.2.3)
Finally, the dimensionless strains may be obtained easily using Eq.
(4.2.3) as well as the dimensionless forms given in Eq. (4.1.7). So, all of
radial displacement, stresses and strains can be completely determined
under the traction conditions on the inner and outer surfaces of the annular
disk. They can be expressed as
). 1 ( at ) 0 ( 0
), 2 . 0 ( at ) 0 ( 0
= = = =
= = = = =
R b r U u
A R a r U u
(4.2.4)
The finite difference approach presented in Chapter 2 is used here to
solve the present problem numerically.
The resolution of the elastic problems of rotating annular disks with
variable thickness are to solve both second-order differential equations
(4.1.11) and (4.1.16) with the aid of the boundary conditions given in Eq.
(4.1.7). Equations (4.1.11) or (4.1.16) can be written in the following
general form:
), ( ) ( ) (
) ( ) ( ) (
R s U R q U R p U
l
l
l
l
l
l
+ + ' = ' ' . 2 , 1 = l (4.3.1)
77
where the prime ) (' denotes differentiation with respect to R and l denotes
the disk number. The functions ) ( , ) ( , ) ( R s R q R p
l l l
may be given by
For annular disk 1:
,
) (
) ( ,
) 2 1 (
) ( ,
) 2 1 (
) (
2
1
2
1 1
k
R
R
R s
R
k
R q
R
k
R p =
+
=
+
=
v
(4.3.2)
For annular disk 2:
,
2
2 1
) (
2
|
|
|
.
|
\
|
+
=
k k
k k k
nA nR
nA nR nA k
e e
e e e knR
R
R p
,
2
2 1
) (
2
2
|
|
|
.
|
\
|
+
=
k k
k k k
nA nR
nA nR nA k
e e
e e e R kn
R
R q
v
. ) (
2
R R s = (4.3.3)
It is clear that the above problems have unique solutions because
), ( ), ( R q R p
l l
and ) (R s
l
are continuous on [0.2, 1] and 0 ) ( > R q
l
on [0.2,
1]. The linear second-order boundary value problems given in (4.3.1)
requires that difference-quotient approximations be used for approximating
) (l
U' and
) (l
U ' ' . First we select an integer 0 > N and divided the interval
[0.2, 1] into ) 1 ( + N equal subintervals, whose end points are the mesh
points , R i R
i
A = for , 1 ..., , 1 , 0 + = N i where ) 1 /( 1 + = A N R . At the interior
mesh points, ,
i
R , ..., , 2 , 1 N i = the differential equation to the approximated
is
), ( ) ( ) ( ) ( ) ( ) (
) ( ) ( ) (
i l i
l
i l i
l
i l i
l
R s R U R q R U R p R U + + ' = ' ' (4.3.4)
If we apply the centered difference approximations of ) (
) (
i
l
R U' and
) (
) (
i
l
R U ' ' to Eq. (4.3.4), we arrive at the system:
76
| |
) (
1
) ( 2 ) (
1
) (
2
1 ) ( ) ( 2 ) (
2
1
l
i
i l
l
i
i l
l
i
i l
U R p
R
U R q R U R p
R
+ (
A
A + +
(
A
+
), ( ) (
2
i l
R s R A = (4.3.5)
for each . ..., , 2 , 1 N i = The N equations, together with the boundary
conditions
. 0
, 0
1
0
=
=
+ N
U
U
(4.3.6)
Are sufficient to determine the unknowns ,
i
U 1 ..., , 2 , 1 , 0 + = N i . The
resulting system of Eq. (4.3.5) is expresses in the tri-diagonal N N -
matrix form:
, B U A = (4.3.7)
where
. 2 , 1 , ..., , 2 , 1 ), ( ) (
, 2 , 1 , 2 , ..., , 4 , 3 , , 2 ..., , 2 , 1 , 0
, 2 , 1 , ..., , 3 , 2 ), (
2
1
, 2 , 1 , 1 ..., , 2 , 1 ), (
2
1
, 2 , 1 , ..., , 2 , 1 ), ( ) ( 2
2
, ,
1 ,
1 ,
2
,
= = A =
= + > = = = =
= =
A
+ =
= =
A
=
= = A =
+
l N i R s R B
l i j N j N i A A
l N i R p
R
A
l N i R p
R
A
l N i R q R A
i l i
i j j i
i l i i
i l i i
i l i i
(4.3.8)
The solutions of the finite difference discretization of the two-point linear
boundary value problems can therefore be found easily even for very small
mesh sizes.
77
The modified Runge-Kutta's method presented in Chapter 2 may be
used here to solve both differential equations (4.1.11) and (4.1.16) with the
aid of the boundary conditions given in Eq. (4.1.7). Equations (4.1.11) or
(4.1.16) can be written in the following general form:
( ) . 2 , 1 , , ,
) ( ) ( ) (
= ' = ' ' l U U R f U
l l
l
l
(4.3.9)
where the prime ) (' denotes differentiation with respect to R and l denotes
the disk number. The functions
l
f may be given by
,
) (
) 2 1 ( ) 2 1 (
2
) 1 (
2
) 1 (
1
k
R
R
U
R
k
U
R
k
f
+
+ '
+
=
v
(4.3.10)
) 2 (
2
2
2 1
U
e e
e e e knR
R
f
k k
k k k
nA nR
nA nR nA k
'
|
|
|
.
|
\
|
+
=
.
2
2 1
) 2 (
2
R U
e e
e e e R kn
R
k k
k k k
nA nR
nA nR nA k
|
|
|
.
|
\
|
v
(4.3.11)
Runge-Kutta's iterative formulae for the second-order differential
equations are
( ),
3
1
) (
3
) (
2
) (
1
) ( ) ( ) (
1
l l l l
i
l
i
l
i
K K K U R U U + + + ' A + =
+
( ). 2 2
3
1
) (
4
) (
3
) (
2
) (
1
) ( ) (
1
l l l l l
i
l
i
K K K K
R
U U + + +
A
+ ' = '
+
(4.3.12)
where R A is the increment of the distance along the radial direction of the
rotating annular disks, and 4 , 3 , 2 , 1 ,
) (
= j K
l
j
can be determined with the
following relations
78
( ), , ,
2
) (
) ( ) (
2
) (
1
l
i
l
i
i l
l
U U R f
R
K '
A
=
,
1
,
4
1
2
1
,
2 2
) (
) (
1
) ( ) (
1
) ( ) (
2
) (
2
|
.
|
\
|
A
+ ' + ' A +
A
+
A
=
l l
i
l l
i
l
i
i l
l
K
R
U K U R U
R
R f
R
K
,
1
,
4
1
2
1
,
2 2
) (
) (
2
) ( ) (
1
) ( ) (
2
) (
3
|
.
|
\
|
A
+ ' + ' A +
A
+
A
=
l l
i
l l
i
l
i
i l
l
K
R
U K U R U
R
R f
R
K
.
2
, ,
2
) (
) (
3
) ( ) (
3
) ( ) (
2
) (
4
|
.
|
\
|
A
+ ' + ' A + A +
A
=
l l
i
l l
i
l
i
i l
l
K
R
U K U R U R R f
R
K
(4.3.13)
The numerical simulation starts from the inner boundary, where a trial
value of the first derivative of the radial displacement is assumed. With a
small distance R A , the radial displacement and its first derivative at the
new position can be obtained using Eq. (4.3.12) and the radial displacement
is calculated. According to the difference between the computed radial
displacement and the known radial displacement at the outer boundary, the
initial trial value of the first derivative of the radial displacement at the
inner boundary is modified and the next iteration is carried out in the same
way. This iterative process is performed until both the boundary conditions
are simultaneously satisfied. Once the radial displacement is obtained, the
stresses and the strains in the rotating annular disks can be obtained using
Eq. (4.1.4) and Eq. (4.1.3) with the aid of the dimensionless given in Eq.
(4.1.7). This process may be easily done after getting the continuous form
of the radial displacement using the curve fitting and least square method.
So, all other quantities can be easily determined.
88
Some numerical examples for the rotating variable-thickness annular
disks will be given according the analytical and numerical solutions
) 3 . 0 ( = v . According to Eq. (4.1.7), the radial displacement, stresses and
strains of the rotating variable-thickness annular disk are determined as per
the analytical solutions are compared with those obtained by the numerical
FDM solutions for disk 1 in Figures 4.34.9. The inner and outer radii of
the disk are taken to be a = 0.2 b (R = A = 0.2) and b (R = 1), and the
results are given in terms of the rotating angular velocity.
The results of the present FDM investigations of the radial displacement
, U for the annular disk 1 with different cases of the geometric parameters
k are reported in Tables 4.1, 4.2, 4.3, 4.4. For these examples, N = 8, 16, 32
and 64, so R A has the corresponding values 0.1, 0.05, 0.025 and 0.0125,
respectively. The FDM gives results compared well with the exact
solution, especially for small values of R A . The relative error between the
exact solution and the FDM one with 0125 . 0 = AR at 6 . 0 = R may be
6
10 61 . 6
for k = 4 . 0 (Table 4.1);
4
10 32 . 2
for k = 4 . 0 (Table 4.2);
5
10 13 . 5
for k = 2 . 0 (Table 4.3); and finally may be
4
10 75 . 1
for k =
2 . 0 (Table 4.4).
88
Table 4.1: Numerical and analytical results for the dimensionless radial
displacement U of a rotating variable-thickness annular disk 1 (k = 4 . 0 )
using finite difference method with different values of R A .
Analytical
FDM
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0 0 0 0 0 0.2000
0.001470751 0.001470384 --- --- --- 0.2125
0.002931746 0.002931115 0.002929253 --- --- 0.2250
0.004384915 0.004384095 --- --- --- 0.2375
0.005831236 0.005830283 0.005827474 0.005816901 --- 0.2500
0.007270953 0.007269912 --- --- --- 0.2625
0.008703742 0.008702645 0.008699407 --- --- 0.2750
0.010128823 0.010127695 --- --- --- 0.2875
0.011545055 0.011543916 0.011540554 0.011527877 0.011486712 0.3000
0.012951008 0.012949872 --- --- --- 0.3125
0.014345014 0.014343893 0.014340587 --- --- 0.3250
0.015725212 0.015724114 --- --- --- 0.3375
0.017089579 0.017088513 0.017085368 0.017073536 --- 0.3500
0.018435963 0.018434932 --- --- --- 0.3625
0.019762100 0.019761109 0.019758187 --- --- 0.3750
0.021065633 0.021064683 --- --- --- 0.3875
0.022344126 0.022343221 0.022340556 0.022330583 0.022299350 0.4000
0.023595080 0.023594219 --- --- --- 0.4125
0.024815936 0.024815121 0.024812728 --- --- 0.4250
0.026004089 0.026003321 --- --- --- 0.4375
0.027156893 0.027156171 0.027154052 0.027146194 --- 0.4500
0.028271668 0.028270991 --- --- --- 0.4625
0.029345700 0.029345068 0.029343218 --- --- 0.4750
0.030376252 0.030375665 --- --- --- 0.4875
0.031360564 0.031360020 0.031358429 0.031352622 0.031336430 0.5000
88
Table 4.1: Continued
Analytical
FDM
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0.032295855 0.032295353 --- --- --- 0.5125
0.033179328 0.033178867 0.033177523 --- --- 0.5250
0.034008172 0.034007750 --- --- --- 0.5375
0.034779561 0.034779177 0.034778064 0.034774101 --- 0.5500
0.035490659 0.035490312 --- --- --- 0.5625
0.036138621 0.036138309 0.036137409 --- --- 0.5750
0.036720592 0.036720314 --- --- --- 0.5875
0.037233710 0.037233464 0.037232760 0.037230376 0.037226336 0.6000
0.037675107 0.037674892 --- --- --- 0.6125
0.038041909 0.038041724 0.038041197 --- --- 0.6250
0.038331238 0.038331080 --- --- --- 0.6375
0.038540210 0.038540078 0.038539711 0.038538618 --- 0.6500
0.038665938 0.038665831 --- --- --- 0.6625
0.038705534 0.038705449 0.038705223 --- --- 0.6750
0.038656102 0.038656040 --- --- --- 0.6875
0.038514750 0.038514707 0.038514603 0.038514506 0.038518179 0.7000
0.038278579 0.038278554 --- --- --- 0.7125
0.037944690 0.037944683 0.037944682 --- --- 0.7250
0.037510184 0.037510192 --- --- --- 0.7375
0.036972158 0.036972180 0.036972266 0.036972873 --- 0.7500
0.036327711 0.036327745 --- --- --- 0.7625
0.035573938 0.035573983 0.035574137 --- --- 0.7750
0.034707936 0.034707991 --- --- --- 0.7875
0.033726800 0.033726863 0.033727066 0.033728093 0.033734890 0.8000
0.032627626 0.032627695 --- --- --- 0.8125
0.031407507 0.031407581 0.031407818 --- --- 0.8250
0.030063540 0.030063617 --- --- --- 0.8375
0.028592817 0.028592898 0.028593149 0.028594317 --- 0.8500
88
Table 4.1: Continued
Analytical
FDM
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0.026992436 0.026992430 --- --- --- 0.8625
0.025259489 0.025259483 0.025259820 --- --- 0.8750
0.023391072 0.023391067 --- --- --- 0.8875
0.021384280 0.021384276 0.021384588 0.021385627 0.021391127 0.9000
0.019236210 0.019236206 --- --- --- 0.9125
0.016943956 0.016943952 0.016944219 --- --- 0.9250
0.014504614 0.014504612 --- --- --- 0.9375
0.011915283 0.011915281 0.011915479 0.011916126 --- 0.9500
0.009173058 0.009173056 --- --- --- 0.9625
0.006275037 0.006275036 0.006275145 --- --- 0.9750
0.003218318 0.003218318 --- --- --- 0.9875
0 0 0 0 0 0.1000
Table 4.2: Numerical and analytical results for the dimensionless radial
displacement U of a rotating variable-thickness annular disk 1 (k = 4 . 0 )
using finite difference method with different values of R A .
Analytical
FDM
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0 0 0 0 0 0.2
0.006318406 0.006319823 0.006324098 0.006341554 0.006416461 0.3
0.014582647 0.014586212 0.014596940 0.014640340 0.014821081 0.4
0.023478567 0.023484234 0.023501276 0.023569983 0.023852901 0.5
0.031163058 0.031170274 0.031191961 0.031279254 0.031636653 0.6
0.035385152 0.035392942 0.035416350 0.035510473 0.035894534 0.7
0.033557488 0.033564501 0.033585569 0.033670231 0.034014924 0.8
0.022804891 0.022809421 0.022823026 0.022877677 0.023099839 0.9
0 0 0 0 0 1.0
88
Table 4.3: Numerical and analytical results for the dimensionless radial
displacement U of a rotating variable-thickness annular disk 1 (k = 2 . 0 )
using finite difference method with different values of R A .
Analytical
FDM
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0 0 0 0 0 0.2
0.010041299 0.010041411 0.010041782 0.010043771 0.010057939 0.3
0.020343564 0.020344437 0.020347087 0.020358162 0.020408370 0.4
0.029553817 0.029555317 0.029559843 0.029578343 0.029657343 0.5
0.036056881 0.036058731 0.036064301 0.036086902 0.036181342 0.6
0.038135919 0.038137804 0.038143474 0.038166400 0.038261175 0.7
0.034022491 0.034024083 0.034028870 0.034048183 0.034127528 0.8
0.021917214 0.021918179 0.021921079 0.021932767 0.021980592 0.9
0 0 0 0 0 1.0
Table 4.4: Numerical and analytical results for the dimensionless radial
displacement U of a rotating variable-thickness annular disk (k = 2 . 0 )
using finite difference method with different values of R A .
Analytical
FDM
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0 0 0 0 0 0.2
0.007405693 0.007407059 0.007411182 0.007428019 0.007500009 0.3
0.016396849 0.016400030 0.016409601 0.016448292 0.016608721 0.4
0.025527401 0.025532162 0.025546471 0.025604118 0.025840498 0.5
0.032944472 0.032950243 0.032967580 0.033037311 0.033321674 0.6
0.036519308 0.036525287 0.036543241 0.036615388 0.036908666 0.7
0.033914044 0.033919238 0.033934831 0.033997454 0.034251489 0.8
0.022622846 0.022626097 0.022635855 0.022675029 0.022833712 0.9
0 0 0 0 0 1.0
88
Table 4.5: Dimensionless radial displacement U of a rotating variable-
thickness annular disk 1 (k = 4 . 0 ) using Richardson extrapolation method
with different values of R A .
Analytical
i 4
Ext
i 3
Ext
i 2
Ext
i 1
Ext
i
R
0 0 0 0 0 0.2
0.011545055 0.011544991 0.011545036 0.011544779 0.011541599 0.3
0.022344126 0.022344073 0.022344109 0.022343880 0.022340995 0.4
0.031360564 0.031360521 0.031360550 0.031360365 0.031358019 0.5
0.037233710 0.037233677 0.037233699 0.037233555 0.037231723 0.6
0.038514750 0.038514725 0.038514742 0.038514635 0.038513281 0.7
0.033726800 0.033726784 0.033726795 0.033726724 0.033725827 0.8
0.021384280 0.021384272 0.021384278 0.021384242 0.021383793 0.9
0 0 0 0 0 1.0
Richardson extrapolation method with , 1 . 0 = AR , 05 . 0 , 025 . 0 and
0.0125 and the obtained results are listed in Table 4.5. These
extrapolations are given, respectively by
,
3
) 1 . 0 ( ) 05 . 0 ( 4
Ext
1
= A = A
=
R U R U
i i
i
(4.4.1a)
,
3
) 05 . 0 ( ) 025 . 0 ( 4
Ext
2
= A = A
=
R U R U
i i
i
(4.4.1b)
,
3
) 025 . 0 ( ) 0125 . 0 ( 4
Ext
3
= A = A
=
R U R U
i i
i
(4.4.1c)
.
15
Ext Ext 16
Ext
1 2
4
i i
i
= (4.4.1d)
Table 4.5 shows that all extrapolations results are correct to the decimal
places listed. In fact, if sufficient digits are maintained, the approximation
87
of
i 4
Ext gives results those agree with the exact solution with maximum
difference error of
6
10 54 . 5
at some of the mesh points.
Now the least square method and curve fitting are used for the discrete
results of the radial displacement U . So, one can get easily the stresses and
strains since we have U as a continuous function of R. The distributions of
the radial displacement, radial and circumferential stresses are presented in
Figure 4.3 for k = 4 . 0 and in Figure 4.4 for k = 4 . 0 . The numerical FDM
solution is compared with the exact analytical solution for the rotating
variable-thickness annular disk 1. It can be seen that the FDM can describe
the radial displacement, stresses and strains through the thickness of the
rotating annular disk very well enough.
86
Figure 4.3: Radial displacement U , radial stress
1
o and circumferential stress
2
o for disk 1, k = 0.4 (Analytical and FDM solutions).
Figure 4.4: Radial displacement U , radial stress
1
o and circumferential stress
2
o for disk 1, k = 4 . 0 (Analytical and FDM solutions).
87
Figure 4.5: Radial displacement U of the variable-thickness annular disk 1 for
two values of k (Analytical and FDM solutions).
Figure 4.6: Radial stress
1
o in the variable-thickness annular disk 1 for two
values of k (Analytical and FDM solutions).
88
Figure 4.7: Circumferential stress
2
o in the variable-thickness annular disk 1
for two values of k (Analytical and FDM solutions).
Figure 4.8: Radial strain
1
c in the variable-thickness annular disk 1 for two
values of k (Analytical and FDM solutions).
888
Figure 4.9: Circumferential strain
2
c in the variable-thickness annular disk 1 for
two values of k (Analytical and FDM solutions).
For the sake of completeness and accuracy, additional results for the
radial displacement, radial stress, circumferential stress, radial strain and
circumferential strain are presented in Figures 4.54.9 for two values of the
geometric parameters k. Figure 4.5 shows the radial displacement U
through the radial direction of the rotating annular disk 1 with k = 2 . 0 .
Similar results for the radial
1
o and the circumferential
2
o stresses are
plotted in Figures 4.6 and 4.7. In addition, similar results for the radial
1
c
and the circumferential
2
c strains are plotted in Figures 4.8 and 4.9. Figure
4.5 shows that the maximum values for the radial displacement U occur at
7 . 0 = R for both k = 2 . 0 and k = 2 . 0 . Figure 4.6 shows that the radial
stress
1
o for k = 2 . 0 is greater than the corresponding one for k = 2 . 0
when 48 . 0 2 . 0 s s R and 1 9 . 0 s s R . However, Figure 4.7 gives
888
circumferential stress
2
o for k = 2 . 0 greater than that of k = 2 . 0 when
64 . 0 2 . 0 s s R and 1 91 . 0 s s R .
Once again, the radial strain
1
c for k = 2 . 0 is greater than the
corresponding one for k = 2 . 0 during 9 . 0 46 . 0 s s R as shown in Figure
4.8. In addition, Figure 4.9 shows that the maximum value of the
circumferential strains
2
c occur at 55 . 0 = R for k = 2 . 0 and at 6 . 0 = R for
k = 2 . 0 .
It is clear that, the FDM gives radial displacement, stresses and strains
consequently, with very good accuracy with the exact analytical solution.
So, it will trustily used to find the solution for Eq. (4.1.16) of the rotating
variable-thickness annular disk 2. As a result, the radial displacement,
stresses and strains in the rotating variable-thickness disk 2 are plotted in
Figures 4.174.22 according to all cases.
Some numerical examples for the rotating variable-thickness annular
disk will be given according the analytical and numerical solutions
) 3 . 0 ( = v . Results determined as per the analytical solutions are compared
with those obtained by the numerical modified Runge-Kutta's method(R-K)
solutions in Figures 4.104.16 for disk 1. The inner and outer radii of the
disk are taken to be a = 0.2 b (R = A = 0.2) and b (R = 1), and the results
are given in terms of the rotating angular velocity
The results of the present investigations for the radial displacement U of
the annular disk 1 are reported in Tables 4.64.9 (k = 4 . 0 and k = 2 . 0 ).
For these examples, N = 8, 16, 32 and 64, so R A has the corresponding
values 0.1, 0.05, 0.025 and 0.0125, respectively. The modified R-K method
gives results compared well with the exact solutions, especially for small
888
values of R A . The relative error between the exact method and the
modified R-K method with 0125 . 0 = AR at 6 . 0 = R may be
7
10 83 . 4
for k = 4 . 0 (Table 4.6);
7
10 36 . 3
for k = 4 . 0 (Table 4.7);
7
10 82 . 5
for k = 2 . 0 (Table 4.8); and finally may be
7
10 64 . 3
for k = 2 . 0
(Table 4.9).
888
Table 4.6: Numerical and analytical results for the dimensionless radial
displacement U of a rotating variable-thickness annular disk 1 (k = 4 . 0 )
using modified Runge Kutta's method with different values of R A .
Analytical
R-K
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0 0 0 0 0 0.2000
0.001470751 0.001470744 --- --- --- 0.2125
0.002931746 0.002931725 0.002931725 --- --- 0.2250
0.004384915 0.004384885 --- --- --- 0.2375
0.005831236 0.005831204 0.005831204 0.005831204 --- 0.2500
0.007270953 0.007270924 --- --- --- 0.2625
0.008703742 0.008703716 0.008703716 --- --- 0.2750
0.010128823 0.010128799 --- --- --- 0.2875
0.011545055 0.011545032 0.011545032 0.011545032 0.011545032 0.3000
0.012951008 0.012950985 --- --- --- 0.3125
0.014345014 0.014344989 0.014344989 --- --- 0.3250
0.015725212 0.015725185 --- --- --- 0.3375
0.017089579 0.017089552 0.017089552 0.017089552 --- 0.3500
0.018435963 0.018435936 --- --- --- 0.3625
0.019762100 0.019762074 0.019762074 --- --- 0.3750
0.021065633 0.021065607 --- --- --- 0.3875
0.022344126 0.022344101 0.022344101 0.022344101 0.022344101 0.4000
0.023595080 0.023595055 --- --- --- 0.4125
0.024815936 0.024815911 0.024815911 --- --- 0.4250
0.026004089 0.026004065 --- --- --- 0.4375
0.027156893 0.027156870 0.027156870 0.027156870 --- 0.4500
0.028271668 0.028271644 --- --- --- 0.4625
0.029345700 0.029345677 0.029345677 --- --- 0.4750
0.030376252 0.030376230 --- --- --- 0.4875
0.031360564 0.031360542 0.031360542 0.031360542 0.031360542 0.5000
888
Table 4.6: Continued
Analytical
R-K
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0.032295855 0.032295834 --- --- --- 0.5125
0.033179328 0.033179308 0.033179308 --- --- 0.5250
0.034008172 0.034008152 --- --- --- 0.5375
0.034779561 0.034779541 0.034779541 0.034779541 --- 0.5500
0.035490659 0.035490640 --- --- --- 0.5625
0.036138621 0.036138602 0.036138602 --- --- 0.5750
0.036720592 0.036720574 --- --- --- 0.5875
0.037233710 0.037233692 0.037233692 0.037233692 0.037233692 0.6000
0.037675107 0.037675090 --- --- --- 0.6125
0.038041909 0.038041893 0.038041893 --- --- 0.6250
0.038331238 0.038331222 --- --- --- 0.6375
0.038540210 0.038540195 0.038540195 0.038540195 --- 0.6500
0.038665938 0.038665924 --- --- --- 0.6625
0.038705534 0.038705519 0.038705519 --- --- 0.6750
0.038656102 0.038656089 --- --- --- 0.6875
0.038514750 0.038514737 0.038514737 0.038514737 0.038514737 0.7000
0.038278579 0.038278566 --- --- --- 0.7125
0.037944690 0.037944678 0.037944678 --- --- 0.7250
0.037510184 0.037510173 --- --- --- 0.7375
0.036972158 0.036972148 0.036972148 0.036972148 --- 0.7500
0.036327711 0.036327701 --- --- --- 0.7625
0.035573938 0.035573928 0.035573928 --- --- 0.7750
0.034707936 0.034707927 --- --- --- 0.7875
0.033726800 0.033726792 0.033726792 0.033726792 0.033726792 0.8000
0.032627626 0.032627618 --- --- --- 0.8125
0.031407507 0.031407500 0.031407500 --- --- 0.8250
0.030063540 0.030063532 --- --- --- 0.8375
0.028592817 0.028592811 0.028592811 0.028592811 --- 0.8500
888
Table 4.6: Continued
Analytical
R-K
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0.026992436 0.026992430 --- --- --- 0.8625
0.025259489 0.025259483 0.025259483 --- --- 0.8750
0.023391072 0.023391067 --- --- --- 0.8875
0.021384280 0.021384276 0.021384276 0.021384276 0.021384276 0.9000
0.019236210 0.019236206 --- --- --- 0.9125
0.016943956 0.016943952 0.016943952 --- --- 0.9250
0.014504614 0.014504612 --- --- --- 0.9375
0.011915283 0.011915281 0.011915281 0.011915281 --- 0.9500
0.009173058 0.009173056 --- --- --- 0.9625
0.006275037 0.006275036 0.006275036 --- --- 0.9750
0.003218318 0.003218318 --- --- --- 0.9875
0 0 0 0 0 0.1000
Table 4.7: Numerical and analytical results for the dimensionless radial
displacement U of a rotating variable-thickness annular disk 1 (k = 4 . 0 )
using modified Runge Kutta's method with different values of R A .
Analytical
R-K
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0 0 0 0 0 0.2
0.009551535 0.009551520 0.009551520 0.009551520 0.009551520 0.3
0.018077905 0.018077889 0.018077889 0.018077889 0.018077889 0.4
0.025116278 0.025116265 0.025116265 0.025116265 0.025116265 0.5
0.029787275 0.029787265 0.029787265 0.029787265 0.029787265 0.6
0.030980716 0.030980708 0.030980708 0.030980708 0.030980708 0.7
0.027406526 0.027406521 0.027406521 0.027406521 0.027406521 0.8
0.017613812 0.017613809 0.017613809 0.017613809 0.017613809 0.9
0 0 0 0 0 1.0
887
Table 4.8: Numerical and analytical results for the dimensionless radial
displacement U of a rotating variable-thickness annular disk 1 (k = 2 . 0 )
using modified Runge Kutta's method with different values of R A .
Analytical
R-K
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0 0 0 0 0 0.2
0.010041299 0.010041270 0.010041270 0.010041270 0.010041270 0.3
0.020343564 0.020343533 0.020343533 0.020343533 0.020343533 0.4
0.029553817 0.029553791 0.029553791 0.029553791 0.029553791 0.5
0.036056881 0.036056860 0.036056860 0.036056860 0.036056860 0.6
0.038135919 0.038135903 0.038135903 0.038135903 0.038135903 0.7
0.034022491 0.034022480 0.034022480 0.034022480 0.034022480 0.8
0.021917214 0.021917208 0.021917208 0.021917208 0.021917208 0.9
0 0 0 0 0 1.0
Table 4.9: Numerical and analytical results for the dimensionless radial
displacement U of a rotating variable-thickness annular disk 1 (k = 2 . 0 )
using modified Runge Kutta's method with different values of R A .
Analytical
R-K
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 1 . 0 = AR
0 0 0 0 0 0.2
0.007405693 0.007405678 0.007405678 0.007405678 0.007405678 0.3
0.016396849 0.016396833 0.016396833 0.016396833 0.016396833 0.4
0.025527401 0.025527387 0.025527387 0.025527387 0.025527387 0.5
0.032944472 0.032944460 0.032944460 0.032944460 0.032944460 0.6
0.036519308 0.036519298 0.036519298 0.036519298 0.036519298 0.7
0.033914044 0.033914037 0.033914037 0.033914037 0.033914037 0.8
0.022622846 0.022622843 0.022622843 0.022622843 0.022622843 0.9
0 0 0 0 0 1.0
886
The numerical applications will be carried out for the radial
displacement, stresses and strains
The distribution of the radial displacement and stresses are presented in
Figure 4.10 for k = 2 . 0 and in Figure 4.11 for k = 2 . 0 . The modified R-K
numerical solution is compared with the exact analytical solution for the
rotating variable-thickness annular disk 1. It is clear that, the modified R-K
method gives all results with excellent accuracy when compared with the
exact analytical solution.
887
Figure 4.10: Radial displacement U , radial stress
1
o and circumferential stress
2
o for disk 1, k = 2 . 0 (Analytical and R-K solutions).
Figure 4.11: Radial displacementU , radial stress
1
o and circumferential stress
2
o for disk 1, k = 2 . 0 (Analytical and R-K solutions).
888
Figure 4.12: Radial displacement U of the variable-thickness annular disk 1 for
two values of k (Analytical and R-K solutions).
Figure 4.13: Radial stress
1
o in the variable-thickness annular disk 1 for two
values of k (Analytical and R-K solutions).
888
Figure 4.14: Circumferential stress
2
o in the variable-thickness annular disk 1
for different values of k (Analytical and R-K solutions).
Figure 4.15: Radial strain
1
c in the variable-thickness annular disk 1 for two
values of k (Analytical and R-K solutions).
888
Figure 4.16: Circumferential strain
2
c in the variable-thickness annular disk 1
for two values of k (Analytical and R-K solutions).
For the sake of completeness and accuracy, additional results for the
radial displacement, radial stress, circumferential stress, radial strain and
circumferential strain are presented in Figures 4.124.16 for two values of
the geometric parameters k. Figure 4.12 shows the radial displacement U
through the radial direction of the rotating annular disk 1 with k = 4 . 0 .
Similar results for the radial
1
o and the circumferential
2
o stresses are
plotted in Figures 4.13 and 4.14. In addition, similar results for the radial
1
c and the circumferential
2
c strains are plotted in Figures 4.15 and 4.16.
Figure 4.12 shows that the maximum values for the radial displacement U
occur at 66 . 0 = R for k = 4 . 0 and at 72 . 0 = R for k = 4 . 0 .
Figure 4.13 shows that the radial stress
1
o for k = 4 . 0 is greater than
the corresponding one for k = 4 . 0 when 92 . 0 47 . 0 s s R . Also, Figure 4.14
shows that the circumferential stresses
2
o for k = 4 . 0 are greater than
those for k = 4 . 0 when 63 . 0 2 . 0 s s R and 1 92 . 0 s s R .
888
However, Figures 4.15 shows that the radial strain
1
c for k = 4 . 0 is
greater than that for k = 4 . 0 when 91 . 0 45 . 0 s s R . Figure 4.16 shows that
the circumferential strain
2
c has two maximum values at 55 . 0 = R and
65 . 0 = R for k = 4 . 0 and k = 4 . 0 , respectively.
It can be seen from Figures 4.104.16 that the modified R-K method
can describe all results through the thickness of the rotating annular disk 1
very well enough. This puts into evidence the great role played by the
modified R-K method in the modeling of rotating variable-thickness
annular disks. So, it will trustily used to find the solution for Eq. (4.1.16) of
the rotating variable-thickness annular disk 2. As a result, the radial
displacement, stresses and strains in the rotating variable-thickness disk 2
are plotted in Figures 4.174.22 according to all cases.
The radial displacement U of the present FDM and modified R-K
method for the rotating variable-thickness annular disk 1 with k = 3 . 0 is
reported in Table 4.10. Additional results for the rotating variable-
thickness annular disk 2 with different cases of the geometric parameters k
and n are reported in Tables 4.114.13. For these examples, R A have the
corresponding values 0.05, 0.025 and 0.0125. The absolute difference error
between the FDM method and the modified R-K method with 0125 . 0 = AR
at 6 . 0 = R may be
7
10 03 . 8
for the rotating variable-thickness annular
disk 1 (Table 4.10).
For the rotating variable-thickness annular disk 2, the absolute
difference errors between the FDM and the modified R-K method with
0125 . 0 = AR at 6 . 0 = R are
6
10 84 . 3
,
6
10 83 . 4
and
6
10 88 . 4
for k
= 2, n = 0.6 (Table 4.11), k = 1.4, n =2 (Table 4.12) and k = 0.8, n = 1.2
(Table 4.13), respectively.
888
Table 4.10: Numerical results for the dimensionless radial displacement
U of a rotating variable-thickness annular disk 1 (k = 3 . 0 ) using finite
difference method and modified Runge-Kutta's method with different
values of R A .
R-K FDM
i
R
0125 . 0 = AR
025 . 0 = AR 05 . 0 = AR 0125 . 0 = AR
025 . 0 = AR 05 . 0 = AR
0 0 0 0 0 0 0.2
0.010781538 0.010781538 0.010781538 0.010781089 0.010779729 0.010774912 0.3
0.021350021 0.021350021 0.021350021 0.021350058 0.021350149 0.021351079 0.4
0.030483888 0.030483888 0.030483888 0.030484393 0.030485886 0.030492326 0.5
0.036684477 0.036684477 0.036684477 0.036685280 0.036687668 0.036697594 0.6
0.038365453 0.038365453 0.038365453 0.038366354 0.038369044 0.038380082 0.7
0.033905194 0.033905194 0.033905194 0.033905993 0.033908380 0.033918114 0.8
0.021665590 0.021665590 0.021665590 0.021666087 0.021667574 0.021673616 0.9
0 0 0 0 0 0 1.0
Table 4.11: Numerical results for the dimensionless radial displacement
U of a rotating variable-thickness annular disk 2 (k = 2, n = 0.6) using finite
difference method and modified Runge-Kutta's method with different
values of R A .
R-K FDM
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 0125 . 0 = AR
025 . 0 = AR 05 . 0 = AR
0 0 0 0 0 0 0.2
0.019725494 0.019725494 0.019725494 0.019720284 0.019704787 0.019644494 0.3
0.032371119 0.032371119 0.032371119 0.032365790 0.032349896 0.032287646 0.4
0.039920734 0.039920734 0.039920734 0.039916071 0.039902138 0.039847362 0.5
0.042650255 0.042650255 0.042650255 0.042646417 0.042634941 0.042589705 0.6
0.040385019 0.040385019 0.040385019 0.040382058 0.040373197 0.040338203 0.7
0.032796136 0.032796136 0.032796136 0.032794095 0.032787987 0.032763822 0.8
0.019483847 0.019483847 0.019483847 0.019482787 0.019479614 0.019467043 0.9
0 0 0 0 0 0 1.0
888
Table 4.12: Numerical results for the dimensionless radial displacement
U of a rotating variable-thickness annular disk 2 (k = 1.4, n = 2) using finite
difference method and modified Runge-Kutta's method with different
values of R A .
R-K FDM
i
R
0125 . 0 = AR
025 . 0 = AR 05 . 0 = AR 0125 . 0 = AR
025 . 0 = AR 05 . 0 = AR
0 0 0 0 0 0 0.2
0.022014588 0.022014588 0.022014588 0.022009059 0.021992604 0.021928343 0.3
0.034902930 0.034902930 0.034902930 0.034896925 0.034879002 0.034808470 0.4
0.042048345 0.042048345 0.042048345 0.042042757 0.042026057 0.041960064 0.5
0.044272218 0.044272218 0.044272218 0.044267385 0.044252931 0.044195665 0.6
0.041592580 0.041592580 0.041592580 0.041588719 0.041577165 0.041531314 0.7
0.033680710 0.033680710 0.033680710 0.033677994 0.033669864 0.033637563 0.8
0.020024225 0.020024225 0.020024225 0.020022803 0.020018544 0.020001610 0.9
0 0 0 0 0 0 1.0
Table 4.13: Numerical results for the dimensionless radial displacement
U of a rotating variable-thickness annular disk 2 (k = 0.8, n = 1.2) using
finite difference method and modified Runge-Kutta's method with different
values of R A .
R-K FDM
i
R
0125 . 0 = AR 025 . 0 = AR 05 . 0 = AR 0125 . 0 = AR
025 . 0 = AR 05 . 0 = AR
0 0 0 0 0 0 0.2
0.021154179 0.021154179 0.021154179 0.021148100 0.021129991 0.021059214 0.3
0.033897407 0.033897407 0.033897407 0.033890965 0.033871729 0.033796064 0.4
0.041231100 0.041231100 0.041231100 0.041225289 0.041207919 0.041139349 0.5
0.043745679 0.043745679 0.043745679 0.043740799 0.043726202 0.043668452 0.6
0.041327629 0.041327629 0.041327629 0.041323827 0.041312452 0.041267374 0.7
0.033588323 0.033588323 0.033588323 0.033585704 0.033577864 0.033546761 0.8
0.020008980 0.020008980 0.020008980 0.020007631 0.020003591 0.019987551 0.9
0 0 0 0 0 0 1.0
888
Figure 4.17: Radial displacement U , radial stress
1
o and circumferential stress
2
o for disk 2 (FDM and R-K solutions).
Figure 4.18: Radial displacement U of the variable-thickness annular disk 2 for
different values of k and n (FDM and R-K solutions).
887
Figure 4.19: Radial stress
1
o in the variable-thickness annular disk 2 for
different values of k and n (FDM and R-K solutions).
Figure 4.20: Circumferential stress
2
o in the variable-thickness annular disk 2
for different values of k and n (FDM and R-K solutions).
886
Figure 4.21: Radial strain
1
c in the variable-thickness annular disk 2 for
different values of k and n (FDM and R-K solutions).
Figure 4.22: Circumferential strain
2
c in the variable-thickness annular disk 2
for different values of k and n (FDM and R-K solutions).
887
Furthermore, the plots of different results for the variable-thickness
annular disk 2 are given in Figures 4.174.22 for various values of k and n.
Figure 4.17 shows that both the modified R-K method and FDM give
accurate results. It is obvious in Figures 4.174.22 that the absolute
difference error between the two methods may be neglected. Figure 4.18
shows that the case of k = 1.4, n = 2 gives the largest displacements while
the case of k = 2, n = 0.6 gives the smallest ones. The difference between
cases may be increases at 6 . 0 = R .
The same discussion may be available for the remainder Figures. The
difference between cases may be increase at 2 . 0 = R (the inner edge) as
show in Figures 4.19 and 4.21. Also it is increasing at 3 . 0 = R as given in
Figure 4.20. Finally, this difference error is increasing at 4 . 0 = R as shown
in Figure 4.22.
This chapter presents analytical and numerical solutions for BVP of
rotating variable-thickness annular disks with a general, arbitrary
configuration. The governing equation is derived from the equilibrium
equation and the stress-strain relationship. The calculation of the rotating
annular disk is turned into finding the solution of a second-order
differential equation under the given conditions at two boundary fixed
points. Finite difference method and modified Runge-Kutta's method are
introduced to solve the governing equation for two types of annular disks,
in which the analytical solution of one of them only is available. A number
of numerical examples is studied. The results from the analytical and
numerical FDM or modified R-K method solutions are compared. The
proposed FDM and modified R-K method approaches give very agreeable
results to the analytical solution. The FDM and modified R-K method
888
presented here may be trustily used for the boundary-value problems that
their analytical solutions are not available.
021
The object of this thesis is to study the rotating of variable-thickness
solid and annular disks. The rotating variable-thickness disks are subjected
to different boundary conditions at the inner and outer edges. The
boundary values problems presented herein are solved for solid and
annular disks at the same angular velocity.
The governing equations of the presented disks are solved analytically
and numerically. The analytical solution is presented with the help of
Whittaker's functions. However the numerical solutions are based upon
both the finite difference method (FDM) as well as the modified Runge-
Kutta's (R-K) method.
Firstly, we discussed the rotating of variable thickness solid disk. The
outer edge is free of prescribed forces and so the radial stress should be
vanished. The results for stress function and the corresponding stresses are
presented in Chapter 3. Additional results for radial displacement and
strains are also presented. The results presented in this chapter are given by
using the analytical and numerical FDM solutions. The FDM solution
gives results compared well with those given by the analytical one except
for minor cases at the outer edge of the solid disk.
Secondly, the rotating variable-thickness annular disk is presented in
Chapter 4. The inner and outer edges of the annular disk are clamped and
so the radial displacement should be vanished. Two annular disks with
variable configuration are discussed. The analytical solution is available for
disk 1. Both the modified R-K method and FDM solutions are also used.
020
They give very accurate results comparing with the analytical solution. The
results are presented in some tables and plotted in some figures.
The analytical solution is not available for dealing with disk 2. So, the
two numerical solutions are used only. In addition, some numerical results
are tabulated and shown graphically using all solution for disk 1. However,
only the numerical solutions are presented graphically for disk 2.
It is to be noted that the relative error between any of the numerical
solutions and the analytical solution is very small. Also the absolute
difference error between the two numerical solutions is very small and may
be neglected. The modified R-K method still gives results more accurate
than the FDM comparing with the analytical solution.
Moreover, the effects of different parameter are also investigated
throughout this thesis. For example, the effect of the geometric parameters
n and k for solid and annular disks is discussed and three cases for different
values of these parameters are studied. These cases show the concavity and
convexity of the solid and annular variable-thickness disks.
211
1. A. S. Thompson: Stresses in rotating discs at high temperature. Trans.
ASME. J. Appl. Mech. Vol. 13, No. 1, pp. 4552 (1946).
2. S. S. Manson: The determination of elastic stresses in gas-turbine
disks. NACA. Report. No. 871, pp. 241251 (1947).
3. S. S. Manson: Direct method of design and stress analysis of rotating
disks with temperature gradient. NACA. Report. No. 952, pp. 103116
(1950).
4. W. R. Leopold: Centrifugal and thermal stresses in rotating disks.
Trans. ASME. J. Appl. Mech. Vol. 15, No. 4, pp. 322326 (1948).
5. S. P. Timoshenko and J. N. Goodier: Theory of Elasticity. 3rd ed.
McGraw-Hill, New York (1970).
6. S. R. Reid: On the influence of acceleration stresses on the yielding of
disks with uniform thickness. Int. J. Mech. Eng. Sci. Vol. 14, No. 11,
pp. 755763 (1972).
7. T. Y. Reddy and H. Srinath: Effect of acceleration stresses on the
yielding of rotating disks. Int. J. Mech. Sci. Vol. 16, pp. 593596
(1974).
8. D. W. A. Rees: The Mechanics of Solids and Structures. McGraw-Hill,
New York (1990).
9. A. C. Ugural and S. K. Fenster: Advanced Strength and Applied
Elasticity. 3rd ed. Prentice-Hall International, London (1995).
10. L. Srinath: Advanced Strength of Materials. Tata McGraw-Hill, New
York (2001).
11. A. N. Sherbourne and D. N. S. Murthy: Stresses in discs with variable
profile. Int. J. Mech. Sci. Vol. 16, No. 7, pp. 449459 (1974).
12. F. Laszlo: Rotating disks in region of permanent deformation. NACA.
TM. No. 1192, pp. 127 (1948).
13. M. B. Millenson and S. S. Manson: Determination of stresses in gas-
turbine disks subjected to plastic flow and creep. NACA. Report. No.
906, pp. 277292 (1948).
14. W. M. H. Lee: Analysis of plane-stress problems with axial symmetry
in strain hardening range. NACA. TM. No. 2217, pp. 179 (1950).
211
15. S. S. Manson: Analysis of rotating discs of arbitrary contour and radial
temperature distribution in the region of plastic deformation. Proc. US.
Natl. Congr. Appl. Mech. Vol. 1, pp. 569577 (1951).
16. W. T. Koiter: On partially plastic thickwalled tubes. In: Biezeno
Anniversary Volume on Applied Mechanics. Haarlem, Holland, pp.
232251 (1953).
17. W. T. Koiter: Stressstrain relations, uniqueness and variational
theorems for elasticplastic materials with a singular yield surface.
Quart. Appl. Math. Vol. 11, pp. 350354 (1953).
18. U. Gamer: Tresca's yield condition and the rotating solid disk. J. Appl.
Mech. Vol. 50, No. 3, pp. 676678 (1983).
19. U. Gamer: Elasticplastic deformation of the rotating solid disk. Ing.
Arch. Vol. 54, No. 5, pp. 345354 (1984).
20. U. Gamer: The rotating solid disk in the fully plastic state. Forsch.
Ing. Wes. Vol. 50, No. 5, pp. 137140 (1984).
21. U. Gamer: Stress distribution in the rotating elasticplastic disk.
ZAMM. Vol. 65, No. 4, pp. T136T137 (1985).
22. U. Gven: Elasticplastic stresses in a rotating annular disk of variable
thickness and variable density. Int. J. Mech. Sci. Vol. 34, No. 2, pp.
133138 (1992).
23. U. Gven: On the applicability of Tresca's yield condition to the linear
hardening rotating solid disk of variable thickness. ZAMM. Vol. 75,
No. 2, pp. 397398 (1995).
24. U. Gven: The fully plastic rotating disk of variable thickness. ZAMM.
Vol. 74, No. 1, pp. 6165 (1994).
25. L. H. You, S. Y. Long and J.J. Zang: Perturbation solution of rotating
solid disks with nonlinear strain-hardening. Mech. Res. Commun. Vol.
24, No. 6, pp. 649658 (1997).
26. L. H. You and J. J. Zhang: Elasticplastic stresses in a rotating solid
disk. Int. J. Mech. Sci. Vol. 41, No. 3, pp. 269282 (1999).
27. U. Gven: Elasticplastic stress distribution in rotating hyperbolic disk
with rigid inclusion. Int. J. Mech. Sci. Vol. 40, No. 1, pp. 97109
(1998).
28. U. Gven: The fully plastic rotating disk with rigid inclusion. ZAMM.
Vol. 77, No. 9, pp. 714716 (1997).
29. D. W. A. Rees: Elasticplastic stresses in rotating discs by Von Mises
and Tresca. ZAMM. Vol. 79, No. 4, pp. 281288 (1999).
211
30. A. N. Eraslan: Inelastic deformations of rotating variable thickness
solid disks by Tresca and von Mises criteria. Int. J. Comp. Eng. Sci.
Vol. 3, No. 1, pp. 89101 (2002).
31. A. N. Eraslan: Von Mises yield criterion and nonlinearly hardening
variable thickness rotating annular disks with rigid inclusion. Mech.
Res. Commun. Vol. 29, No. 5, pp. 339350 (2002).
32. A. N. Eraslan: Thermally induced deformations of campsite tubes
subjected to a nonuniform heat source. J. Therm. Stresses. Vol. 26, pp.
167193 (2003).
33. A. N. Eraslan: Elasticplastic deformations of rotating variable
thickness annular disks with free, pressurized and radially constrained
boundary conditions. Int. J. Mech. Sci. Vol. 45, No. 4, pp. 643667
(2003).
34. A. N. Eraslan and H. Argeso: Limit angular velocities of variable
thickness rotating disks. Int. J. Solids. Struct. Vol. 39, No. 12, pp.
31093130 (2002).
35. A. N. Eraslan and Y. Orcan: On the rotating elasticplastic solid disks
of variable thickness having concave profiles. Int. J. Mech. Sci. Vol.
44, No. 7, pp. 14451466 (2002).
36. A. N. Eraslan and Y. Orcan: Elasticplastic deformation of a rotating
solid disk of exponentially varying thickness. Mech. Mat. Vol. 34, No.
7, pp. 423432 (2002).
37. A. N. Eraslan: Stress distributions in elasticplastic rotating disks with
elliptical thickness profiles using Tresca and von Mises criteria.
ZAMM. Vol. 85, No. 4, pp. 252266 (2005).
38. T. Apatay and A. N. Eraslan: Elastic deformations of rotating
parabolic disks: an analytical solution. J. Fac. Eng. Arch. Gazi. Univ.
Vol. 18, No. 2, pp. 115135 (2003).
39. S. Bhowmick, D. Misra and K. N. Saha: Approximate solution of limit
angular speed for externally loaded rotating solid disk. Int. J. Mech.
Sci. Vol. 50, No. 2, pp. 163174 (2008).
40. A. N. Eraslan, Y. Orcan and U. Gven: Elastoplastic analysis of
nonlinearly hardening variable thickness annular disks under external
pressure. Mech. Research. Commun. Vol. 32, No. 3, pp. 306315
(2005).
41. M. Farshad: Stresses in rotating disks of materials with different
compressive and tensile moduli. Int. J. Mech. Sci. Vol. 16, No. 8, pp.
559564 (1974).
211
42. C. Parmaksizolu and U. Gven: Stresses in rotating nonhomogeneous
disks with different compressive and tensile moduli. I. T. U. J. (J.
Istanbul Teck. Univ. ). Vol. 46, pp. 5862 (1988).
43. U. Gven and C. Parmaksizolu: Tensile and compressive stresses in a
rotating anisotropic circular disk of variable thickness and variable
density. Doa-Tr. J. Eng. Env. Sci. Vol. 15, pp. 517524 (1991).
44. N. Tutuncu: Effect of anisotropy on stresses in rotating discs. Int. J.
Mech. Sci. Vol. 37, No. 8, pp. 873881 (1995).
45. U. Gven, C. Parmaksizolu and O. Altay: Elastic-plastic rotating
annular disk with rigid casing. ZAMM. Vol. 79, No. 7, pp. 499503
(1999).
46. A. M. Zenkour: Thermoelastic solutions for annular disks with
arbitrary variable thickness. Struct. Eng. Mech. Vol. 24, No. 5, pp.
515528 (2006).
47. A. M. Zenkour: Rotating variable-thickness orthotropic cylinder
containing a solid core of uniform thickness. Arch. Appl. Mech. Vol.
76, No. 1-2, pp. 89-102 (2006).
48. A. M. Zenkour: Analytical solutions for rotating exponentially-graded
annular disks with various boundary conditions. Int. J. Struct. Stab.
Dynam. Vol. 5, No. 4, pp. 557577 (2005).
49. A. M. Zenkour and M. N. M. Allam: On the rotating fiber-reinforced
viscoelastic composite solid and annular disks of variable thickness.
Int. J. Comput. Meth. Eng. Sci. Mech. Vol. 7, pp. 21-31 (2006).
50. M. N. M. Allam, R. E. Badr and R. Tantawy: Stresses of a rotating
circular disk of variable thickness carrying a current and bearing a
coaxial viscoelastic coating. Appl. Math. Modelling. Vol. 32, No. 9,
pp. 16431656 (2008).
51. J. N. Reddy, C. M. Wang and S. Kitipornchai: Axisymmetric bending
of functionally graded circular and annular plate. Eur. J. Mech.
A/Solids. Vol. 18, No. 2, pp. 185199 (1999).
52. M. Bayat, M. Saleem, B. B. Sahari, A. M. S. Hamouda and E. Mahdi:
Thermo elastic analysis of a functionally graded rotating disk with
small and large deflections. Thin-Walled Struct. Vol. 45, No. 7-8, pp.
677691 (2007).
53. X. Y. Lee, H. J. Ding and W. Q. Chen: Pure bending of simply
supported circular plate of transversely isotropic functionally graded
material. J. Zhejiang University (Sci. A). Vol. 7, No. 8, pp. 13241328
(2006).
211
54. J. Y. Chen, H.J. Ding and W. Q. Chen: Three-dimensional analytical
solution for a rotating disc of functionally graded materials with
transverse isotropy. Arch. Appl. Mech. Vol. 77, No. 4, pp. 241251
(2007).
55. J. Y. Chen and W. Q. Chen: 3D analytical solution for a rotating
transversely isotropic annular plate of functionally graded materials. J.
Zhejiang University (Sci. A). Vol. 8, No. 7, pp. 10381043 (2007).
56. X.Y. Li, H. J. Ding and W. Q. Chen: Elasticity solutions for a
transversely isotropic functionally graded circular plate subject to an
axisymmetric transverse load
k
qr . Int. J. Solids. Struct. Vol. 45, No.
1, pp. 191210 (2008).
57. M. Ruhi, A. Angoshtari and R. Naghdabadi: Thermoelastic analysis of
thick-walled finite-length cylinders of functionally graded materials. J.
Therm. Stresses. Vol. 28, No. 4, pp. 391408 (2005).
58. Y. Fukui and N. Yamanaka: Elastic analysis for thick-walled tubes of
functionally graded material subjected to internal pressure. JSME. Int.
J. Ser. I. Vol. 35, No. 4, pp. 379385 (1992).
59. Y. Fukui, N. Yamanaka and K. Wakashima|: The stresses and strains
in a thick-walled tube for functionally graded material under uniform
thermal loading. JSME. Int. J. Ser. A. Vol. 36, No. 2, pp. 156162
(1993).
60. J. F. Durodola and O. Attia: Deformation and stresses in FG rotating
disks. Compos. Sci. Technol. Vol. 60, No. 7, pp. 987995 (2000).
61. J. F. Durodola and O. Attia: Property gradation for modification of
response of rotating MMC disks. J. Mater. Sci. Technol. Vol. 16, No.
7-8, pp. 919924 (2000).
62. S. A. H. Kordkheili and R. Naghdabadi: Thermoelastic analysis of a
functionally graded rotating disk. Compos. Struct. Vol. 79, No. 4, pp.
508516 (2007).
63. H. Jahed and S. Sherkatti: Thermoelastic analysis of inhomogeneous
rotating disk with variable thickness. In: Proc. of the EMAS
Conference of Fatigue, Cambridge, England (2000).
64. H. Jahed and R. Shirazi: Loading and unloading behavior of
thermoplastic disk. Int. J. Press. Vess. Piping. Vol. 78, No. 9, pp. 637
645 (2001).
65. B. Farshi, H. Jahed and A. Mehrabian: Optimum design of
inhomogeneous non-uniform rotating disks. Int. J. Comput. Struct.
Vol. 82, No. 9-10, pp. 773779 (2004).
211
66. H. Jahed, B. Farshi and J. Bidabadi: Minimumweight design of
inhomogeneous rotating discs. Int. J. Press. Vess. Piping. Vol. 82, No.
1, pp. 3541 (2005).
67. J. F. Durodola and J. E. Adlington: Functionally graded material
properties for disks and rotors. Key. Eng. Mater. Vol. 127, No. 31, pp.
11991206 (1997).
68. C. O. Horgan and A. M. Chan: The pressurized hollow cylinder or
disk problem for functionally graded isotropic linearly elastic
materials. J. Elast. Vol. 55, No. 1, pp. 4359 (1999).
69. C.O. Horgan and A. M. Chan: The stress response of functionally
graded isotropic linearly elastic rotating disks. J. Elast. Vol. 55, No. 3,
pp. 219230 (1999).
70. S. K. Ha, Y. B. Yoon and S. C. Han: Effect of material properties on
the total stored energy of a hybrid flywheel rotor. Arch. Appl. Mech.
Vol. 70, No. 8-9, pp. 571584 (2000).
71. U. Gven and A. elik: On transverse vibrations of functionally
graded isotropic linearly elastic rotating solid disk. Mech. Res.
Commun. Vol. 28, No. 3, pp. 271276 (2001).
72. A. N. Eraslan and T. Akis: On the plane strain and plane stress
solutions of functionally graded rotating solid shaft and solid disk
problems. Acta. Mech. Vol. 181, No. 1-2, pp. 4363 (2006).
73. L. H. You, X. Y. You, J. J. Zhang and J. Li: On rotating circular disks
with varying material properties. Z. Angew. Math. Phys. Vol. 58, No.
6, pp. 10681084 (2007).
74. F. Vivio and V. Vullo: Elastic stress analysis of rotating converging
conical disks subjected to thermal load and having variable density
along the radius. Int. J. Solids. Struct. Vol. 44, pp. 77677784 (2007).
75. F. Vivio and V. Vullo: Elastic stress analysis of non-linear variable
thickness rotating disks subjected to thermal load and having variable
density along the radius. Int. J. Solids. Struct. Vol. 45, No. 20, pp.
53375355 (2008).
76. A. M. Zenkour: Elastic deformation of the rotating functionally graded
annular disk with rigid casing. J. Mater. Sci. Vol. 42, No. 23, pp.
97179724 (2007).
77. A. M. Zenkour: Stress distribution in rotating composite structures of
functionally graded solid disks. J. Mater. Process. Technol. Vol. 209,
No. 9, pp. 35113517 (2009).
211
78. A. M. Zenkour: Steady-state thermoplastic analysis of a functionally
graded rotating annular disk. Int. J. Struct. Stab. Dynam. Vol. 6, No. 4,
pp. 559-574 (2006).
79. M. Bayat, B. B. Sahari, M. Saleem, A. Ali and S. V. Wong: Bending
analysis of a functionally graded rotating disk based on the first order
shear deformation theory. Appl. Math. Modelling. Vol. 33, No. 11, pp.
42154230 (2009).
80. M. Bayat, B. B. Sahari, M. Saleem, A. Ali and S. V. Wong:
Thermoelastic solution of a functionally graded variable thickness
rotating disk with bending based on the first-order shear deformation
theory. Thin-Walled Struct. Vol. 47, No. 5, pp. 568582 (2009).
81. M. Bayat, M. Saleem, B. B. Sahari, A. M. S. Hamouda and E. Mahdi:
Mechanical and thermal stresses in a functionally graded rotating disk
with variable thickness due to radially symmetry loads. Int. J. Press.
Vess. Piping. Vol. 86, No. 6, pp. 357372 (2009).
82. M. Bayat, M. Saleem, B. B. Sahari, A. M. S. Hamouda and E. Mahdi:
Analysis of functionally graded rotating disks with variable thickness.
Mech. Res. Commun. Vol. 35, No. 5, pp. 283309 (2008).
83. O. C. Zienkiewicz: The Finite Element Method in Engineering
Science. McGraw-Hill, London (1971).
84. P. K. Banerjee and R. Butterfield: Boundary Element Methods in
Engineering Science. McGraw-Hill, New York (1981).
85. L. H. You, Y. Y. Tang, J. J. Zhang and C. Y. Zheng: Numerical
analysis of elasticplastic rotating disks with arbitrary variable
thickness and density. Int. J. Solids. Struct. Vol. 37, No. 52, pp. 7809
7820 (2000).
86. M. H. Hojjati and A. Hassani: Theoretical and numerical analyses of
rotating discs of non-uniform thickness and density. Int. J. Press. Vess.
Piping. Vol. 85, No. 10, pp. 694700 (2008).
87. M. H. Hojjati and S. Jafari: Semi-exact solution of non-uniform
thickness and density rotating disks. Part II: Elastic strain hardening
solution. Int. J. Press. Vess. Piping. Vol. 86, No. 5, pp. 307318
(2009).
88. S. C. Sterner, S. Saigal, W. Kistler and D. E. Dietrich: A unified
numerical approach for the analysis of rotating disks including turbine
rotors. Int. J. Solids. Struct. Vol. 31, pp. 269-277 (1993).
89. M. H. Hojjati and S. Jafari: Semi-exact solution of elastic non-uniform
thickness and density rotating disks by homotopy perturbation and
211
Adomian's decomposition methods. Part I: Elastic solution. Int. J.
Press. Vess. Piping. Vol. 85, No. 12, pp. 871878 (2008).
90. A. M. Zenkour and D. S. Mashat: Analytical and numerical solutions
for a rotating disk of variable thickness. Appl. Meth. Vol. 1, pp. 430
437 (2010).
91. H. S. Martin: Elasticity Theory, Applications, and Numerics. Elsevier,
Amsterdam (2005).
92. M. K. Jain: Numerical Solution of Differential Equations. 2nd ed. John
Wiley & Sons, New York (1984).
93. A. Thom and C. J. Apelt: Field Computations in Engineering and
Physics. D. Van Nostrand, London (1961).
94. P. Linz and R. L. C. Wang: Exploring Numerical Methods: An
Introduction to Scientific Computing Using MATLAB. Jones and
Barteltt Publishers, Boston (2003).
95. J. F. Kenney and E. S. Keeping: Mathematics of Statistics. Pt.1. 3rd
ed. Princeton, NJ, Van Nostrand (1962).
96. F.S. Acton: Analysis of Straight-Line Data. Dover, New York (1966).
97. L. Gonick and W. Smith: The Cartoon Guide to Statistics. Hamper
Perennial, New York (1993).
98. R. L. Burden and J. D. Faires: Numerical Analysis. 8th ed. Thomson,
United Kingdom (2005).
99. A. M. Zenkour and S. A. Al-Ahmadi: Two solutions for the BVP of a
rotating variable-thickness solid disk. Int. J. Natural. Science. Vol. 3,
No. 2, pp. 145153 (2011).
( )
/
- 3122 1432
)
(
( 81 )
.
/
.
.
/
.
/
.
.
.
.
:
.
.
. :
:
- ( ) - ( )
" ."
:
.
" " " " .
" ."
.
:
( " .)"
.
Finite Difference Method
(FDM) .
.
.
.
:
.
.
:
.Finite Difference Method (FDM)
" " " " Modified Runge-Kutta's Method (R-K) .
.
.