Convection-diffusion problems

Q1.
For a 1-D convection diffusion problem, fluid density = 1000kg/m
3
, flow velocity =
1m/s, diffusion coefficient = 10
-9
m
2
/s, and domain length = 1m. Will a central difference
scheme work, for a numerical solution of this problem (Given that dimension of the
solution vector for the TDMA should not exceed 1000)? Give reasons for your answer?
Solution
Cell size
1
m
1000

Cell ( )
3 3
9
9
10 1 10
10 1
10
u
Pe
x

Central difference scheme (CDS) will not work because CDS is suitable for Cell 2 Pe <
Q2.
The temperature variation in condenser tube is given by ( )
0
dT UA
mC T T
dx L
= & , where
m& is the mass flow rate, C is the specific heat, T is the temperature of cooling water,
0
T is
the constant temperature of the condensing steam, U is the overall heat transfer
coefficient, A and is the total heat transfer area. Define a non-dimensional
temperature
0
in
in
T T
T T

,
x
y
L
= , Obtain as a function of y numerically, taking only 5
grid points using upwind scheme. Also compare with the exact solution. You may take
2
AU
mC
=
&

Solution
( )
0 in
T T dT dT d dy d
dx d dy dx L dy

= =

0
0 0
1 1
in
in in
T T T T
T T T T

= =


( ) ( )
0 0
1
in
T T T T =
Governing differential equation becomes
( )
( )( )
0
0
1
in
in
T T d UA
mC T T
L dy L

= &
or, ( ) ( ) 1 2 1
d UA
dy mC

= =
&

or, ( ) ( ) 1 2 1
d UA
dy mC

= =
&

Integrating for an elemental control volume as shown, we get
( ) 2 1
e e
w w
d
dy dy
dy

=


w
e
W P E
Assuming piecewise constant profile for
( ) 2 1
e
e w P
w
dy =

Using upwind scheme
( ) 2 1
P W P
y =
Comparing with the equation
P P E E W W
a a a b = + + , we get
1 2 1.5
P
a y = + = , 0
E
a = , 1
W
a = , 2 0.5 b y = =





From the definition of ,
1
0 =
2 1
1.5 0.5 = +
3 2
1.5 0.5 = +
4 3
1.5 0.5 = +
5 4
1.5 0.5 = +
After solving above equations
2
0.333 =
3
0.555 =
4
0.703 =
5
0.802 =
Exact solution can be found as follows:
2 2
d
dy

+ =
or,
2 2 2
2 2
y y y
d
e e e
dy

+ =
or,
( )
2 2
2
y y
d
e e
dy
=
or,
2 2 y y
e e C = +
At 0 y = , 0 =
0 1 C = +
1 C =
2
1
y
e

=
From the definition of ,
1
0 =
Putting the values of y, we get
2
0.393 = ,
3
0.632 = ,
4
0.777 = ,
5
0.865 =



4
5
3 2 1
0.25

0 0.25 0.5 0.75 1
0
0.2
0.4
0.6
0.8
1
y

Numerical
Analytical

Comparison of numerical solution with analytical solution

Q3.
Consider a 1-D steady state convection diffusion problem without any source term.
Derive a profile assumption for variation of the dependent variable in the advection term,
following the QUICK scheme. ased on that, derive the complete disctretization equation
for the convection-diffusion problem. Assess your discretization in perspective of the
basic rule regarding the sign of coefficients of the discretized equation.
(b) Extend your derivations made in part (a) to a 1 D unsteady state convection-
diffusion problem with fully explicit time discretization.

Solution




2
0 1 2
a a x a x = + +
( ) ( )
2
1 0 1 2 i
a a x a x

= + + (12)
0 i
a = (13)
( )
2
1 0 1 2 i
a a x a x
+
= + + (14)
Subtracting Eq, (14) from Eq. (12) , we get
x=0
i-1 i i+1
1 1 1
2
i i
a x
+
=
or,
1 1
1
2
i i
a
x

+

Adding Eq, (12) and Eq. (14), we have
( )
2
0 2 1 1
2 2
i i
a a x
+
+ = +
or,
( )
1 1
2 2
2
2
i i i
a
x

+
+
=

( )
2
int 0 1 2
2 4
x
x
a a a

= + +
1 1 1 1
2
4 8
i i i i i
i

+ +
+
= + +
1 1
6 3
8
i i i

+
+
=
One-dimensional steady state convection diffusion problem without any source term
can be expressed as
( )
d d d
u
dx dx dx

| |
=
|
\

Integrating with respect to control volume as shown below, we get

e e w w e w
e w
d d
F F
dx dx

= (15)










6 3 6 3
8 8
P E W W P WW
e e w w e w
F F F F


+ + | | | |
=
| |
\ \

1 2 1 2
2 2
2 2
e w e w
e w e w
d d x x
a a a a
dx dx
| | | |
= + +
| |
\ \


2 2
2 2 2 2
E W E W P P WW P WW W
e w
x x x x
+ + | | | |
= + +
| |

\ \

Assuming piecewise linear profiles for the diffusion term, and substituting the same in
Eq. (15), we get
w e
P E
w
u
e
u

W
CV
( ) ( ) ( ) ( ) 6 3 6 3 2 2 2 2
8 8 2 2
e w e w
P E W W P WW E P P W
F F D D
+ + =
Comparing this equation with the standard template equation
P P E E W W WW WW
a a a a = + + , we get
3
8
e
E e
F
a D =
6
8 8
e w
W w
F F
a D = + +
8
w
WW
F
a =
Since
WW
a is negative , the scheme becomes unconditionally unstable.

(b)
( ) ( ) u
t x x x

| |
+ =
|

\

( ) 0 u
t x x

| |
+ =
|

\

Integrating with respect to t and x with piecewise constant profile within a control
volume at a given t for the temporal form
( ) 0
e t t t t e
w t t w
dtdx u dxdt
t x x

+ +
| |
+ =
|

\


( )
0
0
t t
P P
t
e w
x u u dt
x x


+ ( | | | |
+ =
| | (

\ \

( ) ( ) ( ) ( ) ( )
0 0 0 0 0 0 0 0 0 0 0
6 3 6 3 0
8 8
e w
P P P E W e E P W P WW w P W
F F
x D D
( (
+ + + =
( (

Comparing this equation with the standard template equation
0 0
P P E E W W WW WW P P
a a a a a b = + + + + , we get
0
E W WW
a a a = = =
P
x
a
t

0
6 3
8 8
e w
P e w
F F x
a D D
t

= + +

0
0 0 0 0 0
3 6
8 8 8 8
e W w w
E e W W WW w W
F F F
b D D

= + + +
0 0
P P P P
a a b = +