Sharif University of Technology

School of Mechanical Engineering

Center of Excellence in Energy Conversion



Computational Fluid Dynamics

Classification of PDEs


M.T. Manzari
Classification


Classification of PDEs helps us to understand their behavior
and choose suitable solution techniques.
Classification of the governing equations can be performed in
two ways:
o Mathematical
o physical
PDEs are categorized in terms of mathematical
types as follows:
o Hyperbolic (wave)
o Parabolic (transient heat-conduction)
o Elliptic (diffusion)
PDEs are categorize in terms of physical types as
follows:
o Equilibrium Problems (steady heat conduction)
o Marching Problems (Unsteady heat conduction)
Physical Classification

Equilibrium problems involve a closed domain and
boundary conditions.
They are in fact boundary value problems.
Examples are Laplace and Poisson equations:
) . , , (
0
2
2
V =
=
v u f p

D
Boundary condition
PDE must be satisfied in D
Physical Classification

Marching problems involve an open domain (time or
time-like) and initial conditions and/or boundary
conditions.
Example are unsteady heat conduction and boundary
layer (without separation) problems.



x
t
I.C.
B.C.
B.C.
Marching direction
x
y
Physical Classification

Boundary layer equations: (marching in x-direction)
2
2
0
y
u
y
u
v
x
u
u
y
v
x
u

Pure initial value problems:

=
Eq. Berger linear non 0
Eq. Berger linear 0
) ( ) 0 , ( . .
x
u
u
t
u
x
u
a
t
u
x f x u C I


) Eq. wave (
) ( ) 0 , ( & ) ( ) 0 , ( . .
2
2
2
2
2
x
u
a
t
u
x g x u x f x u C I
t

= =




Mathematical Classification

Consider the following general 2
nd
order PDE:


0 = + + + + + + G F E D C B A
y x yy xy xx



The characteristic equation of this PDE is:
0
2
=
|
.
|

\
|

|
.
|

\
|
C
dx
dy
B
dx
dy
A
Whose roots are:
A
AC B B
dx
dy
2
4
2

=
|
.
|

\
|

We can categorize the PDE according to its
discriminant: AC B D 4
2
=
Mathematical Classification...

If 0 > D at ) , (
0 0
y x p , two real characteristics
exist and the equation is Hyperbolic.
If 0 = D at ) , (
0 0
y x p , only one real characteristic
exists and the equation is Parabolic.
If 0 < D at ) , (
0 0
y x p , no real characteristics
exists and the equation is elliptic.
If D changes sign then the PDE is of mixed type.
For example, for 2D potential flow and Transonic
flow, we have:
) 1 ( 4
0 ) 1 (
2
2
M D
M
yy xx
=
= +

Note that a PDE can be of different types in
different regions.

Mathematical Classification...

Wave equation:
) ( ) 0 , ( ), ( ) 0 , ( : . .
,
2
x g x u x f x u C I
u a u
t
xx tt
= =
=

The exact solution is
| |

+

+ + + =
at x
at x
dz z g
a
at x f at x f t x u ) (
2
1
) ( ) (
2
1
) , (
It is seen that
Hyperbolic a D =
2
4
On the t x plane, we can show that two
characteristic lines exist

+ =
+ =
=
2
1
1
c at x
c at x
a dx
dt


Mathematical Classification...

Region 3 is called domain of influence for point P .
Region 2 is called domain (zone) of dependence for
point P .
Region 1, is called domain of silence. This region
does not feel the other two regions.

2
1
1
3
0 0
at x
0 0
at x +
x
t
1/a
-1/a
P
Model Equations

Laplace and Poisson:

= +
) , (
0
y x f
yy xx

Unsteady Heat Conduction:
) (
yy xx t
+ =
First order linear wave equation (Linear Burger Eq.):
0 = +
x t
a
First order nonlinear wave equation (Inviscid nonlinear burger):
0 = +
x t

Viscous Burger Equation:
xx x t
= +
Second order wave equation:
xx tt
a
2
=

System of PDEs

In general, any higher-order PDE can be converted to a
system of first order PDEs.
EXAMPLE: Let us consider the wave equation
xx tt
a
2
= :
0
0
0
2
2
2
2
2
2
=
(

+
(

=
w
v
x a
a
w
v
t
t x
u
a
t
w
x
u
a
x
w
x t
u
x
v
t
v
t
v
x
u
a w
t
u
v

Or equivalently
0
U
A
U
=

x t

Eigen-values of A are a = .
Therefore, we have two distinct real eigen-value.
That is our system is Hyperbolic.
System of PDEs

EXAMPLE: studying the Laplace equation, 0 = +
yy xx
, we
find:
0
0 1
1 0
=
(

+
(

=
v
u
y v
u
x
x
v
y
u

This have two distinct complex eigen-values, i = .
Therefore, the equation is elliptic.
Linear Unsteady System of Equations

Consider
0 = +

y
B
x
A
t

Here, matrices A and B are functions of x t, and y .
is a column vector and is the dependent variable.
is a column vector and is function of x , and y .
The equation is hyperbolic at a point ) , ( t x p if the eigen-values
of A are all real and distinct.
The equation is hyperbolic at a point ) , ( t y p if the eigen-values
of B are all real and distinct.
The equation is parabolic at a point ) , ( t x p if the eigen-values
of A are all real but less than number of equations.
The equation is parabolic at a point ) , ( t y p if the eigen-values
of B are all real but less than number of equations.

Linear Unsteady System of Equations

The equation is elliptic at a point ) , ( t x p if the eigen-values of
A are all complex.
The equation is elliptic at a point ) , ( t y p if the eigen-values of
B are all complex.
The equation is mixed (Hyperbolic/Elliptic) at a point ) , ( t x p if
the eigen-values of A are mixed real and complex.
The equation is mixed (Hyperbolic/Elliptic) at a point ) , ( t y p if
the eigen-values of B are mixed real and complex.






Linear Steady System of Equations

Consider
0 = +

y
B
x
A

Method 1: define
2 3
2 3
4 1
4 1
2
4
b b
a a
b b
a a
R B Q A P
where
PQ R H
+ = = =
=

Then,
Elliptic H
parabolic H
Hyperbolic H
<
=
>
0
0
0

Method 2: re-write the equation as
0 )

( )

( = +

+

j
y
i
x
j B i A
Linear Steady System of Equations

In the characteristic direction n
r
, we can write
y x y x
n B n A n i n B i A T + = + + = ) j

( ) j

(
A wave-like solution will exist if
0 0 = + =
y x
n B n A or T
This gives
0
2
= +
|
|
.
|

\
|
+
|
|
.
|

\
|
P
n
n
R
n
n
Q
y
x
y
x

Or ( PQ R H 4
2
= )
Q
H R
n
n
y
x
2

=
Therefore,
Elliptic H
Parabolic H
Hyperbolic H
<
=
>
0
0
0

Linear Steady System of Equations

EXAMPLE: Steady, inviscid and incompressible flow
0
0
0
=

y
p
y
v
v
x
v
u
x
p
y
u
v
x
u
u
y
v
x
u

In vector form, we have
( ) 1 0 ) (
1 0
0 0
0 1 0
0 0
1 0
0 0 1
0
2 2
= = = + + =
(
(
(

=
(
(
(

=
(
(
(

= = +

y
x
y
x
y x y x
y
U
x
U
n
n
and
v
u
n
n
n n n v n u T
v
v B
u
u A
p
v
u
U with B A

Which means that the equations are of mixed hyperbolic/elliptic
type.
Linear Unsteady System of Equations

EXAMPLE: Unsteady, 1D, inviscid and compressible flow
0
0
1
0
2
=

x
p
u
x
u
a
t
p
x
p
x
u
u
t
u
y
u
x
u
t

In vector form,
(
(
(

=
(
(
(

= =

u u
u
u
p
u with
x t
2
0
/ 1 0
0
0

A U
U
A
U

Whose eigen-values are
a u a u u I + = = , , 0 A
System has 3 real and distinct eigen-values, therefore, it is
hyperbolic.
Second order PDEs

Every second order PDE can be converted to two first order
equations first and then dealt with as before.
EXAMPLE: steady, 2D, viscous, incompressible flow

|
|
.
|

\
|

|
|
.
|

\
|

2
2
2
2
2
2
2
2
Re
1
Re
1
0
y
v
x
v
y
p
y
v
v
x
v
u
y
u
x
u
x
p
y
u
v
x
u
u
y
v
x
u

Set,
b
y
v
x
u
y
u
c
y
v
b
x
v
a
continuum
=

=

Second order PDEs

In vector form, we get
C
y
U
B
x
U
A =

(
(
(
(
(
(
(
(
(

=
(
(
(
(
(
(
(
(
(

=
(
(
(
(
(
(
(
(

=
1 0
Re
1
0 0 0
0
Re
1
0 0 0 0
0 0 1 0 0 0
0 0 0 1 0 0
0 0 0 0 1 0
0 0 0 0 0 1
0 0 0
Re
1
0 0
1 0
Re
1
0 0 0
0 1 0 0 0 0
0 0 1 0 0 0
0 0 0 0 0 1
0 0 0 0 0 0
B A
p
c
b
a
v
u
U

We get,
( )

= +
|
|
.
|

\
|
=
= + =
(
(
(
(
(
(
(
(
(
(

=
0 1
0
0
Re
1
0
Re Re
0 0
Re Re
0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0 0
2 2
2 2 2
x
y
y
y x y
y
y
x
x
y
x
x y
x y
y x
y
n
n
n
n n n T
n
n
n
n
n
n
n n
n n
n n
n
T

Whose eigen-values are imaginary. Therefore, the system is
mixed elliptic/parabolic.