ShihY C Ntut 2009sp AppliedCFD 03BasicOfFiniteDifference

Chapter 3: Basics of Finite Difference
Applied Computational Fluid Dynamics

Y.C. Shih Spring 2009
Chapter 3 Basics of Finite Difference
OUTLINE
3.1 Components of Numerical Methods
3.2 Introduction to Finite Difference
3-3 Errors Involved in Numerical Solutions
3-4 Example
3.1 Components of numerical methods (1)
Properties-1
Consistence
1. The discretization should become exact as the grid spacing
tends to zero
2. Truncation error: Difference between the discretizedequation
and the exact one
Stability: does not magnify the errors that appear in the
course off numerical solution process.
1. Iterative methods: not diverge
2. Temporal problems: bounded solutions
3. Von Neumanns method
4. Difficulty due to boundary conditions and non-linearities
present.
Convergence: solution of the discretizedequations tends to
the exact solution of the differential equation as the grid
spacing tends to zero.
3-1
Conservation
1. The numerical scheme should on both local and global basis respect
the conservation laws.
2. Automatically satisfied for control volume method, either individual
control volume or the whole domain.
3. Errors due to non-conservation are in most cases appreciable only on
relatively coarse grids, but hard to estimate quantitatively
Boundedness:
1. Iterative methods: not diverge
2. Temporal problems: bounded solutions
3. Von Neumanns method
4. Difficulty due to boundary conditions and non-linearities present.
Realizability: models of phenomena which are too complex to treat
directly (turbulence, combustion, or multiphase flow) should be
designed to guarantee physically realistic solutions.
Accuracy: 1. Modeling error 2. Discretization errors 3. Iterative errors
Properties-2
3-2
Discretization methods (Finite Difference)-1
First step in obtaining a numerical solution is to discretize
the geometric domainto define a numerical grid
Each node has one unknown and need one algebraic
equation, which is a relation between the variable value at
that node and those at some of the neighboring nodes.
The approachis to replace each term of the PDE at the
particular node by a finite-difference approximation.
Numbers of equations and unknowns must be equal
3-3
Discretization methods (Finite Difference)-2
Taylor Series Expansion: Any continuous differentiable function, in
the vicinity of x
i
, can be expressed as a Taylor series:
( ) ( ) ( )
( ) ( ) ( )
H
x n
x x
x
x x
x
x x
x
x x x x
i
n
n
n
i
i
i
i
i
i
i i
+

+ +

+ =
!
...
! 3 ! 2
3
3
3
2
2
2
( )
H
x
x x
x
x x
x x x
i
i i
i
i i
i i
i i
i
+

+ +
+
+
3
3
2
1
2
2
1
1
1
6 2
Higher order derivatives are unknown and can be dropped when the
distance between grid points is small.
By writing Taylor series at different nodes, x
i-1
, x
i+1
, or both x
i-1
and
x
i+1
, we can have:
i i
i i
i
x x x

+
+
1
1
1
1

i i
i i
i
x x x
1 1
1 1
+
+

i i
i i
i
x x x
Forward-FDS Backward-FDS
Central-FDS
1
st
order, order of accuracy P
kest
=1
2
nd
order, order of accuracy P
kest
=1
3-4
3-2 Introduction to Finite Difference (1)
Numerical solutions can give answers at only discrete points in the domain,
called grid points.
If the PDEsare totally replaced by a system of algebraic equations which
can be solved for the values of the flow-field variables at the discrete points
only, in this sense, the original PDEshave been discretized. Moreover, this
method of discretizationis called the method of finite differences.
3-5
A partial derivative replaced with a suitable algebraic difference quotient is
called finite difference. Most finite-difference representations of derivatives
are based on Taylors series expansion.
Taylors series expansion:
Consider a continuous function of x, namely, f(x), with all derivatives
defined at x. Then, the value of f at a location can be estimated
from a Taylor series expanded about point x, that is,
In general, to obtain more accuracy, additional higher-order terms must be
included.
x x +
( ) ( ) ... ) (
!
1
...
! 3
1
! 2
1
) ( ) (
3
3
3
2
2
2
+
+ +
+ = +
n
n
n
x
x
f
n
x
x
f
x
x
f
x
x
f
x f x x f
3-6
3-7
Forward, Backward and Central Differences:
(1) Forward difference:
Neglecting higher-order terms, we can get
( )
... ) (
!
) (
... ) (
! 3
) (
! 2
) (
) ( ) ( ) ( ) (
1
3
3
3
1
2
2 2
1
1 1
+

+ +
+ =
+ + +
+ + i
n
n n
i i
i
i i
i
i i
i i i i i
x
f
n
x x
x
f x x
x
f x x
x x
x
f
x f x f
i i i
i
i i
i i
i i
i
x x x
x
x f x f
x x
x f x f
x
f
=
+ +
+
+
+
+
1 1
1
1
1
1
;
) ( ) ( ) ( ) (
) (
(a)
3-8
(2) Backward difference:
Neglecting higher-order terms, we can get
(3) Central difference:
(a)-(b) and neglecting higher-order terms, we can get
( )
... ) (
!
) (
) 1 ( ... ) (
! 3
) (
! 2
) (
) ( ) ( ) ( ) (
1
3
3
3
1
2
2 2
1
1 1
+

+ +
=

i
n
n n
i i n
i
i i
i
i i
i i i i i
x
f
n
x x
x
f x x
x
f x x
x x
x
f
x f x f
1
1
1
1
;
) ( ) ( ) ( ) (
) (

i i i
i
i i
i i
i i
i
x x x
x
x f x f
x x
x f x f
x
f
(b)
1 1
1 1
) ( ) (
) (
+
+
i i
i i
i
x x
x f x f
x
f
(c)
3-9
(4) If , then (a), (b), (c) can be expressed as
Forward:
Backward:
Central:
Note:
x x x
i i
= =
+1
x
f f
x
f
i i
i
+1
) (
x
f f
x
f
i i
i
1
) (
x
f f
x
f
i i
i
+
2
) (
1 1
) (
) (
) (
1 1
1 1

+ +
=
=
=
i i
i i
i i
x f f
x f f
x f f
(d)
(e)
(f)
3-10
Truncation error:
The higher-order term neglecting in Eqs. (a), (b), (c) constitute the
truncation error. The general form of Eqs. (d), (e), (f) plus truncated terms
can be written as
Forward:
Backward:
Central:
) ( ) (
1
x o
x
f f
x
f
i i
i
+
+
) ( ) (
1
x o
x
f f
x
f
i i
i
+

2
1 1
) (
2
) ( x o
x
f f
x
f
i i
i
+
+
3-11
Second derivatives:
* Central difference:
If , then (a)+(b) becomes
* Forward difference:
* Backward difference:
x x x
i i
= =
+1
2
2
1 1
2
2
) (
) (
2
) ( x o
x
f f f
x
f
i i i
i
+
+
=
+
) (
) (
2
) (
2
1 2
2
2
x o
x
f f f
x
f
i i i
i
+
+
=
+ +
) (
) (
2
) (
2
2 1
2
2
x o
x
f f f
x
f
i i i
i
+
+
=

3-12
Mixed derivatives:
* Taylor series expansion:
* Central difference:
* Forward difference:
* Backward difference:
] ) ( , ) [(
! 2
) )( (
2
! 2
) (
! 2
) (
) , ( ) , (
3 3
2
2
2 2
2
2 2
y x o
y x
f y x
y
f y
x
f x
y
f
y
x
f
x y x f y y x x f +

+
+ = + +
] ) ( , ) [(
) )( ( 4
2 2 1 , 1 1 , 1 1 , 1 1 , 1
,
2
y x o
y x
f f f f
y x
f
j i j i j i j i
j i
+

+
=
+ + + +
] ) ( , ) [(
) )( (
, , 1 1 , 1 , 1
,
2
y x o
y x
f f f f
y x
f
j i j i j i j i
j i
+

+
=
+ + + +
] ) ( , ) [(
) )( (
1 , 1 1 , , 1 ,
,
2
y x o
y x
f f f f
y x
f
j i j i j i j i
j i
+

+
=

3-13
Polynomial fitting:
Many applications of polynomial fitting are observed in CFDand heat
transfer. The technique can be used to develop the entire finite-difference
representation for a PDE. However, the technique is perhaps most
commonly employed in the treatment of boundary conditions.
3-14
(1) Consider a second-order polynomial, f(x)=Ax
2
+Bx+C
(a) Select the origin at x
i
, thus,
(b)
(2) For a first-order polynomial f(x)=ax+b
(a)
(b)
x x x x x
i i i
= = =
+ +
2 0
2 1
( ) ( )
( ) ( )

+ + = + + =
+ + = + + =
= = + + =
+ + +
+ + +
C x B x A C Bx Ax f
C x B x A C Bx Ax f
x C C Bx Ax f
i i i
i i i
i i i i
2 2
0
2
2
2
2 2
2
1
2
1 1
2
Q
( )
( )
2
1 2
1 2
2
2
2
3 4
x
f f f
A
x
f f f
B
f C
i i i
i i i
i
+
=
+
=
=
+ +
+ +
+ =
+ =
+ +
b ax f
b ax f
i i
i i
1 1
( ) b x a f
b f
i
i
+ =
=
+1
x
f f
a
f b
i i
i
=
=
+1
x
f f
a
x
f
i i
i

= =
+1
3-15
(3) Finite-difference grid near wall
If we assume that the temperature distribution near the
boundary is a second-degree polynomial of the form
T= a+ by+c y
2
&
Thus, we can evaluate the wall heat flux by the
approximation
b
y
T
y
=
= 0
( )
2
3 2 1 3 2 1
1
2
2
,
) ( 2
4 3
,
y
T T T
c
y
T T T
b T a
+
=
+
= =
( )
) 4 3 (
2
3 2 1
0
T T T
y
k
kb
y
T
k q
y
w
+
= =
=

=
3-16
Alternatively, we can identify the second-degree polynomial as a truncated Taylors
series expansion about y=0.
(1) Second-degree polynomial: T=a+by+cy
2
(2) Taylor series:
(3) Thus, the approximation is equivalent to
utilizing the first three terms of a Taylor series expansion with
the resulting T.E. in the expansion for T being .
Solving the Taylor series for an expression for involves
division by , which reduces the T.E. in the expansion for
to
...
! 3 ! 2
) 0 (
. .
3
0
3
3 2
0
2
2
0
+
+ =
43 42 1
E T
y
y
T y
y
T
y
y
T
T T
2
cy by a T + +
( ) [ ]
3
y o
0
y
T
y
0
y
T
( ) [ ]
2
y o
3-17
The accuracy of changing the mesh size:
(1) To obtain higher resolution in the region where the gradients are
expected to vary rapidly, it is desirable to use a finer mesh over
that particular region rather than refining the mesh over the
entire domain.
(2) Consider a change of the mesh size from to at some
node i.
1
x
2
x
3-18
(3) Approximation of first derivative:
(i) The function f(x) is expanded about the node i in forward and
backward Taylor series, respectively, as
(a)-(b)=>
where means the largest of or
( )
( ) ( )
( ) [ ]
( )
( ) ( )
( ) [ ] ) ( ...
! 3 ! 2
) ( ...
! 3 ! 2
4
1
3
3
3
1
2
2
2
1
1 1
4
2
3
3
3
2
2
2
2
2
2 1
b x o
dx
f d x
dx
f d x
dx
df
x f f
a x o
dx
f d x
dx
f d x
dx
df
x f f
i i
i
i i
i i
i
i i
+
+ =
+
+ + =
+
( ) ( )
( ) [ ] ) ( ...
2
1
2
2
2
1 2
2
1
2
2
1 2
1 1
c x o
dx
f d
x x
x x
x x
f f
dx
df
i
i i
i
+
+

+

=
+
( ) [ ]
2
x o ( ) [ ]
2
1
x o ( ) [ ]
2
2
x o
3-19
(ii) Then the finite-difference approximation of the first derivative
at the node I where the mesh size is changed from to
becomes
Eq. (c) implies eq. (d) is second-order accurate only if
(iii) We note that, if the mesh size varies from to
abruptly, say , then the accuracy of the differencing
at i deteriorates to first-order.
1
x
2
x
) ( ...
1 2
1 1
d
x x
f f
dx
df
i i
i
+

=
+
1 2
x x
1
x
2
x
1 2
2 x x =
3-20
(4) Approximation of second derivative:
(i) (a)+(b)x
=>
where
and means the largest of and
2
1
2
x
x
( ) ( )( ) ( )
( )( ) ( ) [ ] ) ( ...
6
1
1 1
4
3
3
2
2 1 2
2
2
2
2 2
2
1
2
1
e x o
dx
f d
x x x
dx
f d
x
dx
df
x f f f
i
i
i
i i i
+
+ + + + = +
+

( ) [ ]
4
x o
( ) [ ]
4
1
x o ( ) [ ]
4
2
x o
1
2
x
x
=
3-21
(ii) From eq. (e), we can get
The above expression becomes second-order accurate as
and it implies that, unless the mesh spacing is
changed slowly, the truncation error deteriorates.
( )
( )
( ) [ ]
1 2
2
2
2
1
2 2
1
2
2
1 1
x x o
dx
df
x
x
f f f
dx
f d
i
i i i
i
+
+ +
=
+

1 2
x x
3-22
3-3 Errors Involved in Numerical Solutions (1)
In the solution of differential equations with finite differences, a variety of
schemes are available for the discretizationof derivatives and the solution
of the resulting system of algebraic equations.
In many situations, questions arise regarding the round-off and truncation
errorsinvolved in the numerical computations, as well as the consistency,
stability and the convergence of the finite difference scheme.
Round-off errors:computations are rarely made in exact arithmetic. This
means that real numbers are represented in floating point form and as a
result, errors are caused due to the rounding-off of the real numbers. In
extreme cases such errors, called round-off errors, can accumulate and
become a main source of error.
3-23
3-3 Errors Involved in Numerical Solutions (2)
Truncation error: In finite difference representation of derivative with
Taylors series expansion, the higher order terms are neglected by
truncating the series and the error caused as a result of such truncation is
called the truncation error.
The truncation error identifies the difference between the exact solution of
a differential equation and its finite difference solution without round-off
error.
3-24
3-4 Example (1)
3-25
3-4 Example (2)
3-26
3-4 Example (3)
3-27
3-4 Example (4)
3-28
3-4 Example (5)
3-29
3-4 Example (6)
3-30

ShihY C Ntut 2009sp AppliedCFD 03BasicOfFiniteDifference

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ShihY C Ntut 2009sp AppliedCFD 03BasicOfFiniteDifference

Uploaded by

Copyright:

Available Formats

Chapter 3: Basics of Finite Difference

Applied Computational Fluid Dynamics

You might also like