0% found this document useful (0 votes)
2 views4 pages

Modelling

The document outlines a series of problems related to solving partial differential equations (PDEs) including the 1D heat equation with Dirichlet and Neumann boundary conditions, a 2D elliptic equation, and a 1D hyperbolic equation. Each problem involves finding parameters, deriving numerical schemes, ensuring stability, and comparing analytical and numerical solutions. The document also includes specific tasks such as implementing numerical methods and discussing accuracy and errors.

Uploaded by

Madiha Farooq
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
2 views4 pages

Modelling

The document outlines a series of problems related to solving partial differential equations (PDEs) including the 1D heat equation with Dirichlet and Neumann boundary conditions, a 2D elliptic equation, and a 1D hyperbolic equation. Each problem involves finding parameters, deriving numerical schemes, ensuring stability, and comparing analytical and numerical solutions. The document also includes specific tasks such as implementing numerical methods and discussing accuracy and errors.

Uploaded by

Madiha Farooq
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Problem 1: 1D Heat Equation - Dirichlet BC (30%)

Consider the following PDE:



ut = 0.5 · u xx ,
 ( x, t) ∈ [0, 1] × [0, T ]; (the heat equation)
u(0, t) = u(1, t) = 0, t ≥ 0; (boundary condition)

u( x, 0) = −2 sin(4πx ) + 5 sin(9πx ), x ∈ [0, 1]. (initial condition)

(1)

1.1. Find parameters D and ck (k = 1, . . . , n, . . . ) in the formula



∑ ck e−k π Dt sin(kπx),
2 2
u( x, t) =
k =1
so that it represents the solution of (1) (check your answer!).
[5 marks]
1.2. Derive the Forward Differences (FD) explicit scheme for this problem.
[5 marks]
1.3. Define the space and time points x0 = 0, . . . , xm = 1 and t0 = 0, . . . , tn = n∆t = 1.
For a given m ≥ 2, find the values of n ensuring the stability of the method.
[5 marks]
1.4. For a fixed spatial resolution ∆x = 0.01, evaluate and compare results obtained
from Forward difference (FD), Backward differences (BD) and Crank-Nicholson (CN)
numerical schemes. Address the stability and precision of each method.
The analytical solution may be used as a benchmark.
[10 marks]
Hint: For calibration, you could use the graphs:

Figure 1: Numerical and analytical solution u( x, t) at t = 0.05 obtained for the step
values (a) ∆x = 0.1 and ∆t = 0.005 (left); (b) ∆x = 0.05 and ∆t = 0.001 (right).

1.5. Implement a FD scheme solving the problem with the boundary conditions:
u(0, t) = 20 sin 2023 t; u(1, t) = 3t, t ≥ 0.
[5 marks]

3
Problem 2: 1D Heat Equation - Neumann BC (20%)
Consider the following parabolic PDE:

ut = 0.25 · u xx ,
 ( x, t) ∈ [0, 1] × [0, T ]; (the heat equation)
u(0, t) = u x (1, t) = 0, t ≥ 0; (boundary condition)
u( x, 0) = 4 sin 2 x − 3 sin 7π

  
2 x , x ∈ [0, 1]. (initial condition)

(2)

2.1. Find the distinct parameter values A, B, C, D, E in the equation below


 
Bt 5π
u( x, t) = Ae sin x + Ce Dt sin( Ex ),
2

so that it represents the solution of (2) (check your answer!).


[5 marks]

2.2. Write the Forward Difference (FD) formulae for this problem.
[5 marks]

2.3. Illustrate the accuracy and stability of the FD approximation.


[5 marks]

Hint: For calibration, you could use the graph:

Figure 2: Graph of u( x, t) obtained for ∆x = 0.02 and ∆t = 0.002 at t = 25∆t.

2.4. Discuss the absolute and relative error of the numerical method over the grid
defined by ∆x = 0.02 and ∆t = 0.002, for ( x, t) ∈ [0, 1] × [0, 25∆t].
[5 marks]

4
Problem 3: 2D Elliptic Equation (30%)
Consider the following Poisson Equation
u xx ( x, y) + uyy ( x, y) = g( x, y) = 2xy; (3)
( x, y) ∈ D (interior of the contour).

Figure 3: Elliptic Poisson problem with Dirichlet and Neumann BC on a domain D.

3.1. Consider the Dirichlet boundary conditions:


u( x, y) = 1 (on the top/right boundaries);
u(0, y) = y2 , u( x, 0) = x3 . (4)
Using finite differences and ∆x = ∆y = h = 0.2, the discrete vertical and horizontal
coordinates are x1 = 0, . . . , x6 = 1 and y1 = 0, . . . , y6 = 1.
The discretization of the Laplacian is:

ui−1,j + ui+1,j − 4ui,j + ui,j−1 + ui,j+1


(u xx + uyy ) x=xi ,y=y j = = gij = g( xi , y j ). (5)
h2

(a) Write the equations at the coordinates (i, j): (3, 2), (5, 2), (4, 3) and (2, 5).
[6 marks]

(b) Write the system (5) in matrix form.


You could use Au = b, u = (u22 , u32 , u42 , u52 , u23 , u33 , u43 , u24 , u34 , u25 ) T .
[6 marks]

(c) Solve the system for ui,j and plot your results.
[8 marks]

3.2. Solve equation (3) when the lower boundary is insulating:


uy ( x, 0) = 0. (6)
The domain, as well as the lower and upper boundaries are the same. Plot your results.
Hint: Update the system with the extra points u21 , u31 , u41 , u51 .
[10 marks]

5
Problem 4: 1D Hyperbolic Equation (20%)
 2
∂ u ∂2 u


 ∂t 2 = 4 ∂x2
( x, t) ∈ [0, 1] × [0, T ]; (the wave equation)

u(0, t) = u(1, t) = 0; (boundary condition)
(7)


 u( x, 0) = f ( x ) = 3 sin(πx ) − 4 sin3 (πx ); (initial condition 1)
 ∂u( x,0) = g( x ) = 0.

(initial condition 2)
∂t

4.1. Evaluate and discuss the analytical solution of the PDE (7).
[5 marks]
4.2. Solve the PDE numerically using a Forward Differences (FD) explicit scheme.
In what follows the spatial resolution is fixed to ∆x = 0.005.
(a) Write the FD iteration in matrix form and compute the eigenvalues of the matrix.

[2 marks]
(b) Find the maximum allowable time step ∆tr which ensures the stability of FD.

[3 marks]
(c) Illustrate the accuracy and stability of the numerical FD scheme.
[5 marks]

Hint: For calibration, you could use the graph below.

Figure 4: Solution u( x, t) obtained for ∆x = 0.005 and ∆t = ∆tr , evaluated at t = 0.1.

∂u
4.3. Solve (7) exactly and numerically if the second initial condition is ∂t ( x, 0)
= 3.
[5 marks]

You might also like