AE 410 / CSE 461

Computational Aerodynamics
Instructor: Prof. Marco Panesi

1. 1
AE 410
 Lecture 1

Instructor: Prof. Marco Panesi (AE)

Office: 314 Talbot

E-mail: [email protected]
Office hours: 05:00-07:00 pm Wed

Credit: Three hours (CRN: 29954)

Time: Asynchronous
Location: Echo 360

1. 2
AE 410
 Lecture 1

Description (From course catalog) Finite difference and Finite volume

methods for solving partial differential equations. Model equations will be
used to introduce methods prior to application to compressible inviscid
and viscous flows. Emphasis will be placed on developing techniques for
programming and analyzing numerical schemes applicable to more
complicated scenarios.

Prerequisites: Undergraduate level fluid mechanics (AE 312/ ME 310);

introductory numerical methods (AE 370).

Necessary background: Calculus, some differential equations, basic

programming skills.

1. 3
AE 410
 Lecture 1

Textbook: Lomax, Pulliam, & Zingg, Fundamentals of Computational

Fluid Dynamics (PDF freely available on-line)

Other recommended texts: These texts are on reserve in the engineering

library.

• Computational Fluid Mechanics and Heat Transfer by J. Tannehill, D.

Anderson, R. Pletcher; Second Edition, Taylor & Francis.

• Hirsch, Numerical computation of internal and external flows,

Volumes 1 & 2.

• Gustafsson, Kreiss, & Oliger, Time dependent problems and difference

methods (advanced)

• LeVeque, Finite volume methods for hyperbolic problems (advanced)

1. 4
AE 410
 Lecture 1
Grading
Homework/mini-codes: 20%
Midterm: 30%
Project: 50%
Assignments policy: Assignments MUST be returned in timely manner.
Barring extraordinary circumstances, there are no exceptions.

(i) You may work together in study groups, but you must express
comments/conclusions independently.

(i) Homeworks / Take home assignments: are due by the specified date
and time. Late homework papers will only receive 50% credit until the
start of the next class and 0% credit thereafter.
Honor code It is assumed that the UIUC Student Code will be followed
at all times, including during completion of homework and during
exams.

1. 5
AE 410
 Lecture 1

GOAL of CFD:
E
The goal of the Computational Fluid Dynamics (CFD) is to n
numerically solve the governing equation of motion to: g
S
c i
1. Optimize a Design n
i
2. Predict the system performance and response to changes e
e
3. Understand a fluid system in more detail than possible e
n
experimentally. r
c
4. Understand the behavior of the Fluids. i
e
n
g

1. 6
AE 410
 Lecture 1

GOAL of COURSE:

The goal of the Course is to develop solid fundamentals in numerical

methods to be able to:

1. Better understand what a CFD solver does

2. Prepare you for more advanced courses (e.g., TAM 570)
3. Appreciate the Numerical Simulations
4. Encourage you to consider numerical methods as a future
topic of interest.

1. 7
AE 410
 Conservation Laws and Model Equations

We start out by casting our equations in the most general for, the
integral conservation form (useful in understanding Finite Volume):

Z I Z
d ~ · ~n dS =
Q dV + F P dV
dt V (t) S(t) V (t)

• Q is the vector of the conserved variables, i.e. mass, momentum

and energy (per unit volume).
• n is the unit vector normal to the surface (outward).
• F is the set of vector (or tensor) containing the flux of Q (per unit
area and time).
• P is the rate of production per unit volume and time.

1. 8
AE 410
 Conservation Laws and Model Equations

FINITE VOLUME METHOD: Find the solution of Q on the basis

of the various numerical approximations of the integrals above.

The divergence form of the equation can be obtained by applying

Gauss’s Theorem to the flux integral.

@Q ~ ~
+r·F =P
@t

FINITE DIFFERENCE METHOD: Find the solution of Q on the

basis of the various numerical approximations of the differential
operators above.

1. 9
AE 410
 Example – x-Momentum Equation

Let’s start with an example using the x components of the

momentum equation in the integral form.
Z I
d
⇢u dV + ⇢u [~u · n̂] dS =
dt V I S I
p n̂ · êx dS + ⌧¯ n̂ · êx dS
S S

Where u is the x component of the velocity vector u.

Integral form is used in the FINITE VOLUME method.

1. 10
AE 410
 Example – x-Momentum Equation

Let’s start with an example using the x components of the

momentum equation in differential form.

@ ~ · (⇢u~u) = @p ~
(⇢u) + r + r · ~⌧x
@t @x

Where u is the x component of the velocity vector u.

Differential form is used in the FINITE DIFFERENCE method.

1. 11
AE 410
 Example – x-Momentum Equation

It is difficult to start with the full set of equations so let us assume

that:

⇢ = const. ~u = u êx
@p µ = const.
⇡0
@x
2
= µ
3

1. 12
AE 410
 Non-Linear Advection Diffusion Equation

Then the x-momentum equations reduces to:


@u @u 4 @2u
⇢1 +u = µ1
@t @x 3 @x2

Or equivalently
Advection Diffusion

@u @u 4 µ1 @ 2 u
+u =
@t @x 3 ⇢1 @x2

NON-LINEAR advection-diffusion equation.

1. 13
AE 410
 Linearization of the Advection Diffusion Equation

If we go a bit further and we linearize the equation by saying:

u(x, t) = ū + ✏u0 , with ✏⌧1

Where the u is constant and the dependence of x and t comes

through u’. By substituting in the previous equation and
neglecting the higher order terms we get:


@u0 @u0 4 µ1 @ 2 u0
+ ū =
@t @x 3 ⇢1 @x2

LINEAR advection-diffusion equation.

1. 14
AE 410
 Advection and Diffusion Equations

There are two different behaviors in this equation:

@u0 @u0
+ ū = 0, (µ1 = 0)
@t @x
Movement by ū (hyperbolic equation)

0 2 0
✓ ◆
@u @ u 4 µ1
↵ 2 = 0, ū = 0, ↵ = =0
@t @x 3 ⇢1
Diffusion by alpha (elliptic equation)

It turns out that the physical behavior of system has a very strong
influence on the numerics.
1. 15
AE 410
 Model Equations

COMMENT 1: The main focus of the class will be on representative

model equations; the convective and the diffusion equations. These
equations contains many of the salient mathematical and physical features
of the Navier-Stokes and Euler equations.

COMMENT 2: The model equation we study have two properties in

common: they are linear partial differential equations. (see previous eqns)

1. 16
AE 410
 Discrete Representation- 1D space-time

To approximately solve a PDE we need to discretize the domain:

T
i
m
e

a b
X

1. 17
AE 410
 Discrete representation

Uniform grid (or mesh): Uniform spacing between the grid points.

We will set: ✓ ◆
b a
xi = a + ( x) (i 1) where x=
Nx 1
and
✓ ◆
T
tn = t0 + ( t) (n 1) where t=
Nt 1

The approximate solution u(x,t) evaluated at (xi, tn) will be denoted by:
n
ui
1. 18
AE 410
 Numerical Methods for PDEs

Given the advection equation

@u0 @u0
+ ū = 0, (µ1 = 0)
@t @x
In the numerical methods for PDEs there are two main ways to analyze a
given numerical scheme:

Fully Discrete:

✓ n ◆
un+1
i uni uni+1 ui 1
+ ūi =0
t 2 x

1. 19
AE 410
 Numerical Methods for PDEs

Semi-Discrete or Method of Lines:

Treat ui, the solution of u at xi, as a function of time (do NOT discretize in
time).

In this course we will use the method of lines a lot because it will allows us
to write:

1. 20
AE 410
 Method of Lines

Same mathematical form DIFFERENT MATRICES !!!

1. 21
AE 410
 Method of Lines

And because we can often treat M1 and M2 as constant, time invariant

matrices so that if we understand the eigenvalues/eigenvectors of M1 and M2
we understand the numerical method.

1. 22
AE 410
 What have we learned?

1. Introduction to the Course in Computational Aerodynamics

1. Syllabus

1. Finite Difference and Finite Volume discretization

1. Introduction to the model equations and relation with the Navier-Stokes

equations

1. Method of Lines (Brief Introduction)

1. 23
AE 410
 Cancellation

Cancellation: subtraction of nearly equal operands may cause extreme loss of accuracy. When
we subtract two almost equal numbers we set the most significant digits to zero, leaving ourselves
with just the insignificant, and most erroneous, digits. For example, when determining a derivative
of a function the following formula is used:

Intuitively one would want an h very close to zero, however when using floating-point
operations, the smallest number won't give the best approximation of a derivative. As h grows
smaller the difference between f (a + h) and f(a) grows smaller, cancelling out the most significant
and least erroneous digits and making the most erroneous digits more important. As a result the
smallest number of h possible will give a more erroneous approximation of a derivative than a
somewhat larger number. This is perhaps the most common and serious accuracy problem.

1. 24
AE 410