Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Truncation error, consistency, stability and convergence

Truncation error T.E. is the difference between the partial differential


equation (or term) and the difference approximation to it.
T.E. = P.D.E. – F.D.E.
Truncation error contains dissipative error (terms associated with the even
derivative) and dispersive error (terms associated with the odd derivative)

A finite difference representation is said to be consistent if it can be shown


that the truncation error vanishes as the mesh is refined, i.e.

Stability The numerical scheme is stable if the errors (from any sources) are
not permitted to grow in the sequence of numerical procedures from one step to
the next step.
Convergence means that the solution to the F.D.E. approaches the true solution
of the P.D.E.
Consistency + Stability =Convergence
Consistency and stability conditions are the necessary and sufficient conditions
for convergence.
4- Numerical solution of Parabolic Partial Differential Equations
 Equations of motion in fluid mechanics are frequently reduced to parabolic
formulations.
 A typical parabolic second-order PDE is the unsteady one dimensional
incompressible fluid equation, or the unsteady one dimensional heat
conduction equation.
The model equation under consideration has the following form:
𝜕𝑢 𝜕2 𝑢
= ν 2
𝜕𝑡 𝜕𝑦
(one dimensional N.S. equations, incompressible without pressure gradient)
𝜕𝑇 𝜕2 𝑇
=k 2
𝜕𝑡 𝜕𝑦
(Unsteady one dimensional heat conduction equation)

Where “u” is the velocity and “ν” is the kinematic coefficient of viscosity.
“T” is the temperature and “k” is the thermal conductivity.
Using forward finite difference for the time derivative and
explicit central difference for the space derivative (FTCS – 𝜕𝑢 𝜕2 𝑢
= ν 2
Explicit scheme), the parabolic partial differential equation 𝜕𝑡 𝜕𝑦
can be approximated by the following equation:-

This finite difference approximation is a first order in time and second order in
space. For this explicit scheme, the only unknown is “𝑢𝑛+1 ”, and it can be
computed from

“d” is defined as the diffusion number.

To solve this equation in the (t, y) plan, an initial condition and two
boundary conditions must be specified.
Example
Consider the unsteady incompressible flow
between two infinite plates, without
pressure gradient, as shown in figure.
The distance between the two plates is “h”. The lower plate is stationary
while the upper one start to move with velocity “U” in the x-direction at
t=0. Assume that the flow is at rest at “t=0”, h=4cm, U=40 m/s and the
kinematic coefficient of viscosity “ν=0.0002 𝑚2 /s”.
Solution
To start the calculation “Δt and Δy” must be determined.
If the number of grid points in the y-direction is defined as “𝑗𝑚𝑎𝑥 ”
then “Δy=h/(𝑗𝑚𝑎𝑥 -1)”. By selecting Δt, the diffusion number “d” can
be calculated using “d=νΔt/∆𝑦 2 ”.

Assume 𝑗𝑚𝑎𝑥 =5 and Δt=0.1sec Δy=0.01 m and d=0.2.


t =0 t=0.1 t=0.2 t=0.3 t=0.4 t=0.5 t=∞
sec sec sec sec sec sec (s.s)
y j n=1 n=2 n=3 n=4 n=5 n=6 n=∞
0.04 5 40 40 40 40 40 40 40
0.03 4 0 8 12.8 16 18.304 20.058 30
0.02 3 0 0 1.6 3.52 5.376 7.066 20
0.01 2 0 0 0 0.32 0.896 1.613 10
0 1 0 0 0 0 0 0 0
Assume 𝑗𝑚𝑎𝑥 =5 and Δt = 5 sec Δy=0.01 m and d=1

t =0 t=0.5 t=1.0 t=1.5 t=2.0 t=2.5 t=∞


sec sec sec sec sec sec (s.s)
y J n=1 n=2 n=3 n=4 n=5 n=6 n=∞
0.04 5 40 40 40 40 40 40 40
0.03 4 0 40 0 80 -80 280 ?
0.02 3 0 0 40 -40 160 -320 ?
0.01 2 0 0 0 40 -80 240 ?
0 1 0 0 0 0 0 0 0
Numerical experiments show that if d > 0.5 the FTCS-Explicit
scheme solution is unstable
Others explicit scheme for the parabolic model equation:-

1. Richardson scheme

2.DuFort-Frankelscheme
FTCS – Implicit scheme of the parabolic model equation is

This finite difference approximation is a first order in time and second order
in space and is known as the Laasonen scheme.

For this implicit scheme, the unknowns are “𝑢𝑛+1 ”, and it can be computed
after inversion of a tridiagonal matrix as

“d” is defined as the diffusion number.

To solve this equation in the (t, y) plan, an initial condition and two boundary
conditions must be specified.
The same previous example
Assume 𝑗𝑚𝑎𝑥 =5 and Δt=0.5 sec Δy=0.01 m and d=1.0.
2
- 𝑛+1
𝑢𝑗−1 + 3 𝑢𝑗𝑛+1 - 𝑢𝑗+1
𝑛+1
= 𝑢𝑗𝑛 At n=1 - 𝑢𝑗−1 + 3 𝑢𝑗2 - 𝑢𝑗+1
2
= 𝑢𝑗1

- 𝑢12 + 3 𝑢22 - 𝑢32 = 𝑢21


- 𝑢22 + 3 𝑢32 - 𝑢42 = 𝑢31
- 𝑢32 + 3 𝑢42 - 𝑢52 = 𝑢41
40
t =0 sec t=0.5 t=1.0 t=1.5 t=2.0 t=2.5 (s.s)
sec sec sec sec sec
y J n=1 n=2 n=3 n=4 n=5 n=6 n=∞
0.04 5 40 40 40 40 40 40 40
0.03 4 0 15.238 21.95 25.315 27.169 28.255 30
0.02 3 0 5.714 10.612 13.994 16.193 17.595 20
0.01 2 0 1.905 4.172 6.056 7.416 8.337 10
0 1 0 0 0 0 0 0 0

Numerical experiments show that the FTCS-Implicit scheme


solution is unconditional stable

You might also like