Programming in applied fluids

Chapter 2

Finite Different Methods (FDM)

Dr. Ngo Ich Long

School of Transportation Engineering
Hanoi University of Science and Technology (HUST)
27-Oct-23 1
 Contents
Chapter 02: Finite Different Methods

2.1. Approximation in FDM

2.2. Finite different method
2.3. Difference schemes
2.4. Applying FDM for 1D convection equation

27-Oct-23 2
 2.1. Approximation in FDM
* Practice in MATLAB: Roots of a Quadratic

27-Oct-23 3
 2.1. Approximation in FDM
* Practice in MATLAB: Roots of a Quadratic

27-Oct-23 4
 2.1. Approximation in FDM
* Practice in MATLAB: Roots of a Quadratic

27-Oct-23 5
 2.1. Approximation in FDM
* Practice in MATLAB: the parachutist problem

Practice-01:

27-Oct-23 6
 2.1. Approximation in FDM
* Practice in MATLAB: the parachutist problem

Practice-02: Using M-file → analpara.m

27-Oct-23 7
 2.1. Approximation in FDM
* Practice in MATLAB: the parachutist problem

Practice-03: Using M-file → analpara2.m

27-Oct-23 8
 2.1. Approximation in FDM
* Practice in MATLAB: the parachutist problem

Practice-04:

27-Oct-23 9
 2.1. Approximation in FDM
* Practice in MATLAB: the parachutist problem

Practice-04: Using M-file → numpara.m

27-Oct-23 10
 2.1. Approximation in FDM
* Taylor series
+ In essence, the Taylor series provides a means to predict a function value at one
point in terms of the function value and its derivatives at another point.
+ In particular, the theorem states that any smooth function can be approximated as a
polynomial.

Zero-order approximation
First-order approximation

Second-order approximation
………………………..

Taylor series expansion

27-Oct-23 11
 2.1. Approximation in FDM
* Remainder in Taylor series

+ Remainder term:

+ It is often convenient to simplify Taylor series with h=xi+1 – xi

or

27-Oct-23 Graphical depiction of a zero-order Taylor series 12

prediction and remainder.
 2.1. Approximation in FDM
* An example for Taylor series

Exact estimate
27-Oct-23 13
 2.1. Approximation in FDM
* The Effect of Nonlinearity and Step Size on the Taylor Series Approximation

Solution:

Effect of step size h:

27-Oct-23 14
 2.2. Finite different method
* Introduction
+ In numerical analysis, finite-difference methods (FDM) are a class of numerical
techniques for solving differential equations by approximating derivatives with
finite differences.
+ FDM converts ordinary differential equations (ODE) or partial differential
equations (PDE), which may be nonlinear, into a system of linear equations that
can be solved by matrix algebra techniques.
+ Today, FDM are one of the most common
approaches to the numerical solution of
PDE, along with others like FEM, FVM,…

27-Oct-23
A practical example of Burger equation 15
 2.2. Finite different method
* Introduction

27-Oct-23 16
 2.2. Finite different method
* Basic idea in FDM

27-Oct-23 17
 2.3. Difference schemes
* Let’s start with forward difference scheme

Based on Taylor series expansion, we have

f ( xi +1 ) = f ( xi ) + f ' ( xi )( xi +1 − xi ) + Rn
Neglecting truncation Rn , we have

f ( xi +1 ) = f ( xi ) + f ' ( xi )( xi +1 − xi )
f ( xi +1 ) − f ( xi ) fi
f ' ( xi ) = =
xi +1 − xi h

Note: It is termed a “forward” difference

because it utilizes data at i and i + 1 to
estimate the derivative.

27-Oct-23 18
 2.3. Difference schemes
* Three schemes of first-order difference approximation
Forward difference Downwind scheme
f ( xi +1 ) − f ( xi )
method
fi
f ' ( xi ) = =
xi +1 − xi h

Backward difference Upwind scheme

method
f ( xi ) − f ( xi −1 ) fi
f ' ( xi ) = =
xi − xi −1 h

Center difference Center-scheme

f ( xi +1 ) − f ( xi −1 ) f ( xi +1 ) − f ( xi −1 )
method
f ' ( xi ) = =
xi +1 − xi −1 2h

All the above versions can also be developed for second,

third, and higher derivatives. 19
27-Oct-23
 2.3. Difference schemes
* Second-order difference schemes
Forward difference Downwind scheme
method
f ( xi + 2 ) − f ( xi +1 ) f ( xi +1 ) − f ( xi )
− f ( xi + 2 ) − 2 f ( xi +1 ) + f ( xi )
f "( xi ) = h h =
h h2
Backward difference Upwind scheme
method
f ( xi ) − f ( xi −1 ) f ( xi −1 ) − f ( xi − 2 )
− f ( xi ) − 2 f ( xi −1 ) + f ( xi − 2 )
f " ( xi ) = h h =
h h2

Center difference Center-scheme

method
f ( xi +1 ) − f ( xi ) f ( xi ) − f ( xi −1 )
− f ( xi +1 ) − 2 f ( xi ) + f ( xi −1 )
f "( xi ) = h h =
h h2

All the above versions can also be developed for second,

third, and higher derivatives. 20
27-Oct-23
 2.3. Difference schemes
* Three schemes of difference approximation

Forward difference Backward difference Center difference

The Taylor series analysis yields the practical information

that the centered difference is a more accurate
representation of the derivative.

27-Oct-23 21
 2.3. Difference schemes
* An example

Estimate first derivative of function at x=0.5 using h=0.5 and 0.25

Solution:

For both step sizes, the centered difference approximation

is more accurate than forward or backward differences. 22
27-Oct-23
 2.3. Difference schemes
* For two-dimensional problem

By using these approximations to replace partial derivatives, the PDEs are

converted into difference equations and the resultant system of algebraic
27-Oct-23 equations are solved using any direct or iterative methods. 23
 2.3. Difference schemes
* Explicit and implicit methods

 u u − u l l  u uil++11 − uil +1
 x = x
i +1 i
 x = x
 2  2 l +1 l +1 l +1
 l
− l
+ l
  u u − 2u + u
 u
=
ui +1 2ui ui −1 = i +1 i i −1

 x 2 x 2  x 2 x 2
27-Oct-23 24
 2.4. Applying FDM for 1D convection equation
* Theory
u u
+c =0 1D convection/advection equation, also called wave Eq.
t x
where u(x,t), x Є R, is scalar (wave) advected by a nonzero constant c during time t.
+ Sign of c characterizes the direction of wave propagation and magnitude of c show how
fast the wave propagates.
+ If initial value for equation is given as u0(x) = u(x,0), the exact solution of equation is:

u ( x, t ) = u0 ( x − ct )

27-Oct-23 25
 2.4. Applying FDM for 1D convection equation
* Implementation

Computational domain with ghost nodes

+ For transient term, uin+1 − uin uin − uin−1

u uin+1 − uin +c =0
= t x
t t ct n
+ For convection term with
n +1
→ ui = ui − n

x
( ui − uin−1 )
backward difference ct
CFL = is called Courant/convection number
u uin − uin−1 x
=
x x Stability criteria: CFL <= 1
27-Oct-23 26
 2.4. Applying FDM for 1D convection equation
* Solution in Matlab

N=50, ∆t=0.009
→ CFL = 0.45

N=110, ∆t=0.009
→ CFL = 0.99

27-Oct-23 27
 2.4. Applying FDM for 1D convection equation

N=110, ∆t=0.01
→ CFL = 1.1

N=110, ∆t=0.009
→ CFL = 0.99

N=110, ∆t=0.003
→ CFL = 0.33

27-Oct-23 28
 2.1. Approximation in FDM
* Basic Idea

27-Oct-23 30
 2.1. Approximation in FDM
* A simple mathematical model

where F = net force acting on the body (N, or kg m/s2), m

= mass of the object (kg), and a = its acceleration (m/s2)

(1)

More advanced techniques must be applied to obtain an

exact or analytical solution.

27-Oct-23 31
 2.1. Approximation in FDM
* A simple mathematical model

Substituting into Eq. (1) to give

Euler’s method
27-Oct-23 32