Scribd
Upload
90 views

Boundary-Value Ordinary Differential Equations: CE 601: Numerical Methods

The document summarizes boundary value ordinary differential equations (BV-ODEs) and the equilibrium finite difference method used to solve them. Some key points: 1) BV-ODEs involve dependent variables that vary with respect to space and often arise from equilibrium problems in closed domains. 2) The equilibrium finite difference method discretizes the spatial domain into grid points and applies the ODE and finite difference approximations at each point to generate a system of algebraic equations to solve. 3) Examples are provided for solving BV-ODEs with different boundary conditions like Dirichlet, Neumann, and mixed boundary conditions using this method.

Uploaded by

Amlan Shome
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
90 views

Boundary-Value Ordinary Differential Equations: CE 601: Numerical Methods

The document summarizes boundary value ordinary differential equations (BV-ODEs) and the equilibrium finite difference method used to solve them. Some key points: 1) BV-ODEs involve dependent variables that vary with respect to space and often arise from equilibrium problems in closed domains. 2) The equilibrium finite difference method discretizes the spatial domain into grid points and applies the ODE and finite difference approximations at each point to generate a system of algebraic equations to solve. 3) Examples are provided for solving BV-ODEs with different boundary conditions like Dirichlet, Neumann, and mixed boundary conditions using this method.

Uploaded by

Amlan Shome
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 17

CE 601: Numerical Methods

Lecture 28

Boundary-Value Ordinary
Differential Equations

Course Coordinator:
Dr. Suresh A. Kartha,
Associate Professor,
Department of Civil Engineering,
IIT Guwahati.
• As suggested, based on the auxiliary
conditions we can define ODE as:
o IV-ODE
o BV-ODE
• Some examples of BV-ODE:
• Beam bending problem

d4y
EI q ( x ); y (0) 0, y ( L) 0, y "(0) 0, y "( L) 0
dx 4
y deflection (dependent variable)
• Heat diffusion in a steel rod:

d 2T 2 2
T Ta
dx 2

T x1 T1
T x2 T2

• The dependent variable varying w.r.t space


generally leads to boundary-value ODEs.
• Boundary value ODEs are often found in
equilibrium problems and in closed domains.
• Following types of BV-ODEs are generally
encountered:
– Single Boundary Value ODE
– System of Boundary Value ODEs
– Linear and Non-linear ODEs
– Homogeneous and Non-homogeneous ODEs
• A general second-order non-linear BV-ODE is:
d2y dy
P x, y Q x, y y F ( x ); y ( x1 ) y1
dx 2 dx
y ( x2 ) y2

• A general second order linear BV-ODE is:


d2y dy
P Qy F ( x ); y ( x1 ) y1
dx 2 dx
y ( x2 ) y2
• Solution domain of the above equations is
x1 x x2
• Boundary conditions for BV-ODEs are:

– Dirichlet Boundary condition e.g. y( x1 ) c1

dy
– Neumann Boundary condition e.g. c2
dx x1

dy
– Mixed Boundary condition e.g. ay(x1 ) b c3
dx x1
The Equilibrium Finite-Difference Method
• We will use the equilibrium finite difference
method to solve the boundary-value ODEs.
– Discretize the entire continuous spatial domain
into smaller discrete grid points.
– Apply the ODE at each grid point.
– Approximate the derivatives at any grid point
using finite difference formulas.
– Obtain the corresponding finite-difference
algebraic equations.
– Solve them
• Consider the following 1-D second-order
linear BV-ODE:
d2y dy
P( x) Q( x) y F ( x); y ( x0 ) y0 , y ( xL ) yL
dx 2 dx

• The domain has been discretized.


• At any general node i, we have
d2y dy
P ( xi ) Q( xi ) yi F ( xi ) (1)
dx 2 i
dx i

• Now, d2y yi 1 2 yi yi 1
; O( x 2 )
dx 2 i x2
dy yi yi
1 1
; O( x 2 )
dx i 2 x

• Substituting in (1), we have


yi 1 2 yi yi 1 yi 1 yi 1
Pi Qi yi Fi
x2 2 x
• Rearranging the terms, we have
1 Pi 2 1 Pi
yi 1 yi Qi yi 1 Fi
x2 2 x x2 x2 2 x
or
x 2 x
1 Pi yi 1 2 x Qi yi 1 Pi yi 1 Fi x 2 (2)
2 2

Equation (2) is the finite-difference equation


for grid node ‘i’.
This equation is applied to each grid node ‘i’ to
generate a system of algebraic equations.
• Example: Solve the heat transfer BV-ODE for a
rod of length 1.0 cm. The governing ODE is:
d 2T 2 2
T Ta
dx 2
2
where 16.0 cm -2 (heat diffusivity)
Ta 20 C(ambient temperature)
The B.Cs T (0.0) 25 C
T (1.0) 75 C
• Solution: The rod is of length = 1.0 centimeter (cm). Let
us discretise the length as follows:
x 0.25 cm. As per the given data T0 250 C and T4 750 C. T1 , T2 , T3 are unknowns.
d 2T
For any node: 16Ti 16 20 320
dx 2 i

Ti 1 2Ti Ti 1
i.e. 2
16Ti 320
x
i.e. Ti 1 2Ti Ti 1 Ti 20
i.e. Ti 1 3Ti Ti 1 20 (1)
Eq.(1) is applied at unknown nodes 1,2,3.
At i 1, At i 2, At i 3,
T2 3T1 T0 20 T1 3T2 T3 20 T2 3T3 T4 20
T2 3T1 25 20 T2 3T1 95
3T1 T2 45
3 1 0 T1 45
i.e. 1 3 1 T2 20
0 1 3 T3 95
Solve this system of linear equations to get T1 , T2 , T3 .
• To solve problems with Neumann B.C.s
d2y dy dy
P x Q x y F ( x); y ( x0 ) y0 , m
dx 2 dx dx x L
• The FDE (2) is applied in each node ‘i’:
x x
1 Pi yi 1 2 x 2Qi yi 1 Pi yi 1 Fi x 2
2 2

• At i N , Neumann B.C. is given.


• Consider an imaginary node N+1,
i.e. not physically present and is used only for
computational purpose.
dy yN 1 yN 1
m m yN 1 yN 1 2m x
dx xN 2 x
• Substitute the previous expression for yN 1 at
node i N (FDE)
x x
i.e. 1 PN yN 1 2 x 2QN yN 1 PN y N 1 FN x 2
2 2
x x x
i.e. 1 PN 1 PN yN 1 2 x 2QN yN FN x 2 1 PN 2m. x
2 2 2

or, 2 yN 1 2 x 2QN yN FN x 2 2m. x m. x 2 PN (3)

• The FDE (3) is to be applied at the boundary


node, where Neumann B.C. is given.
In the above-mentioned example
dT
if 0 and T ( xL ) 75 C are the boundary conditions,
dx x 0

then the FDE is to be applied in nodes i 0,1, 2, 3.


dT T1 T 1
For i 0, 0 0 T1 T1
dx x 0 2 x
At i 0, T1 3T0 T1 0 20
3T0 2T1 0 0 20
At i 1, T0 3T1 T2 0 20
At i 2, 0 T1 3T2 T3 20
At i 3, 0 0 T2 3T3 95
3 2 0 0 T0 20
1 3 1 0 T1 20
System of linear equations:
0 1 3 1 T2 20
0 0 1 3 T3 95
• To solve Mixed B.C. problems:
• If the mixed B.C. is given at one-end for the
general linear second order (1-D) BV-ODE
dy
Ay (x N ) B p
dx xN

• Then at node i=N, the FDE


yN 1 yN 1
AyN B p
2 x
or, B yN 1 yN 1 2 p x 2 A xy N
1
or, yN 1 yN 1 2 p x 2 A xy N
B
Substitute this expression of yN 1 in FDE (2).

You might also like