BCHE201L: Computational Methods in Chemical Engineering

Dr. Shishir Kumar Behera

School of Chemical Engineering (SCHEME)
VIT Vellore
 Parabolic Equations

HEAT-CONDUCTION EQUATION

• Heat balance for the differential element in the long,

thin insulated rod.
• Heat balance also considers the amount of heat stored in
the element over a unit time period Dt
• Input - Output = Storage

Dr. S. K. Behera BCHE201L Computational Methods in Chemical Engineering 2

• Substituting Fourier’s law of heat conduction,

result in

• Parabolic equations can be solved by substituting finite

divided differences for the partial derivatives.
• While elliptic equations were bounded in all relevant
dimensions, parabolic PDEs are temporally open-ended.

Dr. S. K. Behera BCHE201L Computational Methods in Chemical Engineering 3

 Parabolic Equations
A second order linear PDE (2 - independen t variable s x , y )
A u xx  B u xy  C u yy  D  0,
A, B, and C are functions of x and y
D is a function of x, y, u, u x , and u y

is parabolic if B 2  4 AC  0

4
 Parabolic Problems

 T ( x, t )  2 T ( x, t )
Heat Equation : 
t x 2
T (0, t )  T (1, t )  0
T ( x,0)  sin( x )

* Parabolic problem ( B 2  4 AC  0)
* Boundary conditions are needed to uniquely specify a solution.

5
 Solution of the Heat Equation
• Two solution techniques to the Parabolic Equation
(Heat Equation) are presented:
1. Explicit Method:
Simple, Stability Problems.
2. Crank-Nicolson Method:
Involves the solution of a Tridiagonal system of
equations, Stable.

6
 Explicit Method

T ( x, t )  2T ( x, t )

t x 2
T ( x, t  k )  T ( x, t ) T ( x  h, t )  2T ( x, t )  T ( x  h, t )

k h2
T ( x, t  k )  T ( x, t )  2 T ( x  h, t )  2T ( x, t )  T ( x  h, t ) 
k
h
k
Define   2
h
T ( x, t  k )   T ( x  h, t )  (1  2  ) T ( x, t )   T ( x  h, t )

7
 Explicit Method
How do we compute?

T ( x, t  k )   T ( x  h, t )  (1  2  ) T ( x, t )   T ( x  h, t )
means

T(x,t+k)

T(x-h,t) T(x,t) T(x+h,t)

8
 Convergence and Stability of the Solution

• Convergence
The solutions converge means that the solution
obtained using the finite difference method approaches
the true solution as the steps x and t approach zero.

• Stability:
An algorithm is stable if the errors at each stage of the
computation are not magnified as the computation
progresses.

9
 Convergence and Stability of the Solution

T ( x, t  k ) can be computed directly using :

T ( x, t  k )   T ( x  h, t )  (1  2  ) T ( x, t )   T ( x  h, t )

Can be unstable errors are magnified 

1 h2
To guarantee stability, (1  2  )  0     k 
2 2
This means that k is much smaller th an h
This makes it slow.

10
Example 1:
Solve the PDE :
 2u(x,t) u(x,t)
 0
 x
2
t
u(0, t )  u(1, t )  0
u( x,0)  sin( x )

Use h  0.25, k  0.25 to find u( x, t ) for x  [0,1], t  [0,1]

k
  2 4
h

11
  2 u ( x, t )  u ( x, t )
 0
x 2
t
u ( x  h , t )  2u ( x , t )  u ( x  h , t ) u ( x , t  k )  u ( x , t )
2
 0
h k
16u( x  h, t )  2u( x, t )  u( x  h, t )   4u( x, t  k )  u( x, t )   0

u ( x , t  k )  4 u ( x  h, t )  7 u ( x , t )  4 u ( x  h, t )

12
 u( x, t  k )  4 u( x  h, t )  7 u( x, t )  4 u( x  h, t )

t=1.0 0 0

t=0.75 0 0

t=0.5 0 0
t=0.25 0 0

t=0 0 0
Sin(0.25π) Sin(0. 5π) Sin(0.75π)

x=0.0 x=0.25 x=0.5 x=0.75 x=1.0

13
u(0.25,0.25)  4 u(0,0)  7 u(0.25,0)  4 u(0.5,0)
 0  7 sin( / 4)  4 sin( / 2)  0.9497

t=1.0 0 0

t=0.75 0 0

t=0.5 0 0
t=0.25 0 0

t=0 0 0
Sin(0.25π) Sin(0.5π) Sin(0.75π)

x=0.0 x=0.25 x=0.5 x=0.75 x=1.0

u( x, t  k )  4 u( x  h, t )  7 u( x, t )  4 u( x  h, t )
14
u(0.5,0.25)  4 u(0.25,0)  7 u(0.5,0)  4 u(0.75,0)
 4 sin( / 4)  7 sin( / 2)  4 sin( 3 / 4)  0.1716

t=1.0 0 0

t=0.75 0 0

t=0.5 0 0
t=0.25 0 0

t=0 0 0
Sin(0.25π) Sin(0. 5π) Sin(0.75π)

x=0.0 x=0.25 x=0.5 x=0.75 x=1.0

15
 Remarks

The obtained results are probably not accurate

because : 1  2  7

For accurate results : 1  2  0

h 2 (0.25) 2
One needs to select k    0.03125
2 2
k
For example, choose k  0.025, then   2  0.4
h

16
u( x, t  k )  0.4 u( x  h, t )  0.2 u( x, t )  0.4 u( x  h, t )

t=0.10 0 0

t=0.075 0 0

t=0.05 0 0
t=0.025 0 0

t=0 0 0
Sin(0.25π) Sin(0. 5π) Sin(0.75π)

x=0.0 x=0.25 x=0.5 x=0.75 x=1.0

17
u(0.25,0.025)  0.4 u(0,0)  0.2 u(0.25,0)  0.4 u(0.5,0)
 0  0.2 sin( / 4)  0.4 sin( / 2)  0.5414

t=0.10 0 0

t=0.075 0 0

t=0.05 0 0
t=0.025 0 0

t=0 0 0
Sin(0.25π) Sin(0. 5π) Sin(0.75π)

x=0.0 x=0.25 x=0.5 x=0.75 x=1.0

18
u(0.5,0.025)  0.4 u(0.25,0)  0.2 u(0.5,0)  0.4 u(0.75,0)
 0.4 sin( / 4)  0.2 sin( / 2)  0.4 sin( 3 / 4)  0.7657

t=0.10 0 0

t=0.075 0 0

t=0.05 0 0
t=0.025 0 0

t=0 0 0
Sin(0.25π) Sin(0. 5π) Sin(0.75π)

x=0.0 x=0.25 x=0.5 x=0.75 x=1.0

Repeat the procedure to find the values at other points

19
Practice Problem
Solve the heat equation subject to the conditions u(x,0)
= x2(25-x2), u(0,t) = u(5,t) = 0. With h = 1 and k = 0.5, compute
the value of uij for i = 0,1,2,3,4,5 and j = 0,1,2 using Explicit
formula.

20
 Crank-Nicolson Method

• This method involves solving a Trigonal system of linear

equations.

• The method is stable (No magnification of error)

• We can use larger h, k (compared to the explicit method)

21
Crank-Nicolson Method

22
 Crank-Nicolson Method
 2 u ( x, t )  u( x, t )
Heat Equat ion :  becomes
x 2
t

u ( x  h , t )  2u ( x , t )  u ( x  h , t ) u ( x , t )  u ( x , t  k )
2

h k
k
2
u( x  h, t )  2u( x, t )  u( x  h, t )   u( x, t )  u( x, t  k )
h
k k k
 2 u( x  h, t )  (1  2 2 ) u( x, t )  2 u( x  h, t )  u( x, t  k )
h h h

23
 Crank-Nicolson Method
k
Define   2 then Heat equation becomes :
h
  u( x  h, t )  (1  2 ) u( x, t )   u( x  h, t )  u( x, t  k )

u(x-h,t) u(x,t) u(x+h,t)

u(x,t - k)

24
 Crank-Nicolson Method
The equation :
  u( x  h, t )  (1  2 ) u( x, t )   u( x  h, t )  u( x, t  k )
can be rewritten as :
  ui 1, j  (1  2 ) ui , j   ui 1, j  ui , j 1
and can be expanded as a system of equations (fix j  1) :
  u0,1  (1  2 ) u1,1   u2,1  u1,0
  u1,1  (1  2 ) u2,1   u3,1  u2,0
  u2,1  (1  2 ) u3,1   u4,1  u3,0
  u3,1  (1  2 ) u4,1   u5,1  u4,0

25
 Crank-Nicolson Method
  u( x  h, t )  (1  2 ) u( x, t )   u( x  h, t )  u( x, t  k )
can be expressed as a Tridiagona l system of equations :
1  2    u1,1   u1,0   u0,1 
   1  2   u   u 
   2,1    2, 0 
   1  2    u3,1   u3,0 
  u  u   u 
   1  2   4,1   4,0 5,1 

where u1,0 , u2,0 , u3,0 , and u4,0 are the initial temperatu re values
at x  x0  h, x0  2h, x0  3h, and x0  4h
u0,1 and u5,1 are the boundary values at x  x0 and x0  5h

26
 Crank-Nicolson Method
The solution of the tridiagon al system produces :
The temperatu re values u1,1, u2,1, u3,1, and u4,1 at t  t0  k
To compute the temperatu re values at t  t0  2k
Solve a second tridiagon al system of equations ( j  2)
1  2    u1,2   u1,1   u0,2 
   1  2   u   u 
   2, 2    2,1 
   1  2   u3,2   u3,1 
     
   1  2   u
 4,2   4,1u   u5, 2 

To compute u1,2 , u2,2 , u3,2 , and u4,2

Repeat the above step to compute temperatu re values at t0  3k , etc.
27
Example 2
Solve the PDE :
 2u( x, t ) u( x, t )
 0
 x
2
t
u(0, t )  u(1, t )  0
u( x,0)  sin( x )

Solve using Crank - Nicolson method

Use h  0.25, k  0.25 to find u( x, t ) for x  [0,1], t  [0,1]

28
  2 u ( x, t )  u ( x, t )
 0
x 2
t
u ( x  h , t )  2u ( x , t )  u ( x  h , t ) u ( x , t )  u ( x , t  k )
2

h k
16u( x  h, t )  2u( x, t )  u( x  h, t )   4u( x, t )  u( x, t  k )   0
k
Define   2  4
h
 4 u ( x  h, t )  9 u ( x , t )  4 u ( x  h, t )  u ( x , t  k )
 4 ui 1, j  9 ui , j  4 ui 1, j  ui , j 1

29
 4u0,1  9u1,1  4u2,1  u1,0  9u1,1  4u2,1  sin( / 4)
 4u1,1  9u2,1  4u3,1  u2,0  4u1,1  9u2,1  4u3,1  sin( / 2)
 4u2,1  9u3,1  4u4,1  u3,0   4u2,1  9u3,1  sin( 3 / 4)

u1,4 u2,4 u3,4

t4=1.0 0 0
u1,3 u2,3 u3,3
t3=0.75 0 0
u1,2 u2,2 u3,2
t2=0.5 0 0
u1,1 u2,1 u3,1
t1=0.25 0 0

t0=0 0 0
Sin(0.25π) Sin(0.5π) Sin(0.75π)
x0=0.0 x1=0.25 x2=0.5 x3=0.75 x4=1.0

30
 Solution of First Row at t2= 0.25 sec

The Solution of the PDE at t1  0.25 sec is the solution

of the following tridiagon al system of equations :
 9 4   u1,1  sin( 0.25 )
 4 9  4 u    sin( 0.5 ) 
   2,1   
  4 9  u3,1  sin( 0.75 )
 u1,1   0.21151 
  
 u2,1   0.29912 
 
u3,1   0.21151 

31
 Second Row at t2= 0.5 sec
 4u0,2  9u1,2  4u2,2  u1,1  9u1,2  4u2,2  0.21151
 4u1,2  9u2,2  4u3,2  u2,1  4u1,2  9u2,2  4u3,2  0.29912
 4u2,2  9u3,2  4u4,2  u3,1   4u2,2  9u3,2  0.21151
u1,4 u2,4 u3,4
t4=1.0 0 0
u1,3 u2,3 u3,3
t3=0.75 0 0
u1,2 u2,2 u3,2
t2=0.5 0 0
u1,1 u2,1 u3,1
t1=0.25 0 0

t0=0 0 0
Sin(0.25π) Sin(0. 5π) Sin(0.75π)
x0=0.0 x1=0.25 x2=0.5 x3=0.75 x4=1.0

32
 Solution of Second Row at t2= 0.5 sec

The Solution of the PDE at t2  0.5 sec is the solution

of the following tridiagon al system of equations :
 9 4   u1,2   u1,1   0.21151 
 4 9  4 u   u   0.29912 
   2,2   2,1   
  4 9  u3,2  u3,1   0.21151 
 u1,2  0.063267 
  
 u2,2   0.089473 
 
u3,2  0.063267 

33
 Solution of Row 3 at t3= 0.75 sec

The Solution of the PDE at t3  0.75 sec is the solution

of the following tridiagon al system of equations :
 9 4   u1,3   u1,2  0.063267 
 4 9  4 u   u   0.089473 
   2,3   2, 2   
  4 9  u3,3  u3,2  0.063267 
 u1,3  0.018924 
 
 u2,3   0.026763 
 
u3,3  0.018924 

34
 Solution of Row 4 at t4= 1 sec

The Solution of the PDE at t4  1 sec is the solution

of the following tridiagon al system of equations :
 9 4   u1,4   u1,3  0.018924 
 4 9  4 u   u   0.026763 
   2, 4   2,3   
  4 9  u3,4  u3,3  0.018924 
 u1,4  0.0056606 
 
 u2,4   0.0080053 
 
u3,4  0.0056606 

35
 Remarks
The Explicit Method:
• One needs to select small k to ensure stability.
• Computation per point is very simple but many points are needed.

Cranks Nicolson:
• Requires the solution of a Tridiagonal system.
• Stable (Larger k can be used).

36
Practice Problem
Solve the heat equation subject to the conditions u(x,0)
= 0, u(0,t) = 0 and u(1,t) = t. With h = 0.25 and k = 0.0625,
compute the value of u(0.5, 0.125) using Crank-Nicolson formula.

37
 A SIMPLE IMPLICIT METHOD
• Implicit methods overcome both these difficulties at the
expense of somewhat more complicated algorithms.

38
Comparison

• In implicit methods, the spatial derivative is approximated at an advanced time level l +1.
For example, the second derivative would be approximated by

39
 A SIMPLE IMPLICIT METHOD
The fundamental difference between explicit and implicit
approximations is depicted here. For the explicit form,
we approximate the spatial derivative at time level l
(Fig.a). Recall that when we substituted this
approximation into the partial differential equation, we
obtained a difference equation with a single unknown
Til+1 .Thus, we can solve “explicitly” for this unknown. In
implicit methods, the spatial derivative is approximated
at an advanced time level l+1. For example, the second.
derivative would be approximated by (Fig. b).

Computational molecules demonstrating the fundamental

differences between (a) explicit and (b) implicit methods.