0% found this document useful (0 votes)

138 views

Solution Algorithm

The document discusses various solution algorithms used in computational fluid dynamics (CFD), including: - Difference equations which represent partial derivatives as finite difference quotients, allowing partial differential equations to be represented as algebraic equations. - Explicit and implicit approaches for solving the resulting difference equations, with explicit being conditionally stable and implicit being unconditionally stable but requiring solving a system of equations. - Numerical methods like the Lax-Wendroff method and MacCormack's method which use Taylor series expansions to march the solution in time until steady state is reached. - The alternating direction implicit technique which splits the equations into alternating directions to reduce computation. - Iterative relaxation techniques to solve systems of

Uploaded by

MEET PATEL

Available Formats

Download as PPTX, PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

138 views

Solution Algorithm

Uploaded by

MEET PATEL

Available Formats

Download as PPTX, PDF, TXT or read online on Scribd

You are on page 1/ 94

Computational Fluid

Dynamics
Solution algorithms

1-1

Difference Equation
In CFD we represent the partial derivatives by means of algebraic finite
difference quotient. When all the partial derivatives in partial difference
equation is replace by e finite difference quotient then resulting algebraic
equation is known as difference equation.
Consider unsteady, one dimension heat conduction equation with constant
thermal diffusivity,

It is a parabolic partial difference equation. As such, it lends it self to a

marching solution with respect to time

1-2

Difference Equation

1-3

Difference Equation

1-4

Difference Equation

1-5

Difference Equation

1-6

Explicit and Implicit approach

1-7

Explicit and Implicit approach

1-8

Explicit and Implicit approach

1-9

Explicit and Implicit approach

1-10

Explicit and Implicit approach

1-11

Explicit and Implicit approach

1-12

Explicit and Implicit approach

1-13

Explicit and Implicit approach

1-14

Explicit and Implicit approach

1-15

Explicit and Implicit approach

1-16

Explicit and Implicit approach

1-17

Explicit and Implicit approach

1-18

Explicit and Implicit approach

1-19

Explicit and Implicit approach

1-20

Explicit and Implicit approach

1-21

Explicit and Implicit approach

1-22

Example

1-23

Example

1-24

Example

1-25

Example

1-26

Example

1-27

Example

1-28

Example

1-29

The LaxWendroff Method

Let us describe this method by considering a simple gas-dynamic problem,
namely the subsonicsupersonic isentropic flow through a convergentdivergent
nozzle, as sketched in Fig. 1

Here, a nozzle of specified area distribution, A = A(x), is given, and the reservoir
conditions are known. Let us consider a quasi-one-dimensional solution where
the flow field variables are functions of x (in the steady state).
1-30

The LaxWendroff Method

The nozzle is divided into a number of grid points in the x-direction as shown in
Fig. 1; the spacing between adjacent grid points is x.
Now assume values of the flow field variables at all grid points, and consider
this rather arbitrarily assumed flow as an initial condition at time t = 0
In general, these assumed values will not be the exact steady-state results;
indeed, the exact steady-state results are what we are trying to calculate.
Consider a grid point, say point i. Let gi denote a flow field variable at this point
(gi might be pressure, density, velocity, etc.). This variable g i will be a function
of time; however, we know gi at time t = 0, i.e. we know gi(0) because we have
assumed values for all the flow field variables at all the grid points at the initial
time t = 0.
We now calculate a new value of gi at time t +t; starting from the initial
conditions, the first new time is t+t = 0+t. Here, t is a small increment in
time to be discussed later. The new value of g i, i.e. gi(t+t), is obtained from a
1-31
Taylors series expansion in time as

The LaxWendroff Method

1-32

The LaxWendroff Method

1-33

The LaxWendroff Method

1-34

The LaxWendroff Method

1-35

The LaxWendroff Method

1-36

The LaxWendroff Method

The behaviour of this type of solution is illustrated in Figs. 7.2. In Fig. 7.2, the
temperature distribution through a given nozzle is shown. The dashed line
labelled t = 0 is the initially assumed values for T throughout the nozzle.
The curve above it labelled 8t is the temperature distribution after eight time
steps following the above procedure. The curves labeled 16t and 32t are
similar results after 16 and 32 time steps respectively. Note that the temperature
1-37
distribution has rapidly changed from the assumed initial distribution
at t = 0.

The LaxWendroff Method

At later times, the changes become smaller; note that the curve labelled 120t is
not too different from that for 32t. Finally, after 744 time steps, the changes are
so small that the temperature distribution is essentially at a steady state. This
steady state is the desired solution.
Fig. 7.3 illustrates the variation of mass flow, m, through the nozzle. The dashed
line is the m consistent with the assumed initial conditions at t=0. The curves
labelled 16t and 32t graphically demonstrate the wild variations in m at early
times.
However, after 120 time steps m has become more stable, and after 744 time steps
has reached a steady state. This steady state distribution for m is a straight,
horizontal line, as it should be for steady flow, where m=constant through the
nozzle
The method described above, utilizing Eq. (7.1), which is the first three terms of a
Taylors series expansion and where both the first and second partial derivatives in
Eq. (7.1) are found by finite-differencing the spatial derivatives in the governing
1-38
flow equations with central differences, is called the Lax-Wendroff
method.

The LaxWendroff Method

1-39

MacCormacks Method

MacCormacks method, has been the most popular explicit finite-difference

method for solving fluid flows. It is closely related to the Lax-Wendroff method,
but is easier to apply.

Let us use the same nozzle problem discussed in previous Section to illustrate
MacCormacks method in the present section. MacCormacks method, like the
Lax-Wendroff method, is based on a Taylors series expansion in time. Once
again, as in previous Section, let us consider the density at grid point i.

1-40

MacCormacks Method

1-41

MacCormacks Method

1-42

MacCormacks Method
The above predictorcorrector approach is carried out for all grid points
throughout the nozzle, and is applied simultaneously to the momentum and energy
equations in order to generate uit+t and eit+t.
In this fashion, the flow field through the entire nozzle at time t +t is calculated.
This is repeated for a large number of time steps until the steady state is achieved,
just as in the case of the Lax-Wendroff method described in previous section.
MacCormacks technique as described above, because a two-step predictor
corrector sequence is used with forward differences on the predictor and rearward
differences on the corrector, is a second-order accurate method. Therefore, it has
the same accuracy as the Lax-Wendroff method described in previous section
However, the MacCormack method is much easier to apply, because there is no
need to evaluate the second time derivatives as was the case for the Lax-Wendroff
method. To see this more clearly, recall Eqs. (7.7) and (7.8), which are required for
the Lax-Wendroff method but MacCormacks method does not require such
second derivatives, and hence does not deal with equations such as Eqs. (7.7) and
1-43
(7.8).

MacCormacks Method
A few comments are made with regard to the specific application to the quasi
one- dimensional nozzle flow shown in Fig. 7.1.
At the inflow boundary (the first grid point at the left), the values of p, T and
are fixed, independent of time, and are assumed to be reservoir values. The inflow
velocity, which is a very small subsonic value, is calculated from linear
extrapolation using the adjacent internal points, or it can be evaluated from the
momentum equation applied at the first grid point using one-sided differences.
At the outflow boundary (the last grid point at the right in Fig. 7.1), all the
dependent variables are obtained from linear extrapolation from the adjacent
internal points, or by applying the governing equations at this point, using onesided differences
Finally, we note that results obtained from the LaxWendroff method and from
the MacCormack method are virtually identical

1-44

Stability Criterion
Examine Eq. (7.1), which is vital to the LaxWendroff method. Note that it
requires the specification of a time increment, t. Examine Eqs. (7.9) and (7.11),
which are vital to the MacCormack method. They too require the specification of a
time increment, t.
For explicit methods, the value of t cannot be arbitrary, rather it must be less
than some maximum value allowable for stability. The time-dependent
applications described in Sects. 7.2 and 7.3 are dealing with governing flow
equations which are hyperbolic with respect to time. For a two or threedimensional flow

Where C is the Courant number. The maximum allowable value of C for stability in
1-45 from approximately
explicit time-dependent finite difference calculations can vary

ADI (Alternating Direction Implicit) Technique

Consider a unsteady two dimensional heat conduction equation

In implicit (Crank Nicolson) form it can be written as,

1-46

ADI (Alternating Direction Implicit) Technique

1-47

ADI (Alternating Direction Implicit) Technique

1-48

ADI (Alternating Direction Implicit) Technique

1-49

ADI (Alternating Direction Implicit) Technique

1-50

ADI (Alternating Direction Implicit) Technique

1-51

ADI (Alternating Direction Implicit) Technique

1-52

ADI (Alternating Direction Implicit) Technique

1-53

ADI (Alternating Direction Implicit) Technique

1-54

ADI (Alternating Direction Implicit) Technique

1-55

Relaxation Technique (Iterative technique)

1-56

Relaxation Technique (Iterative technique)

Grid point 1 to 20 are boundary node, whose value is known to us

based on boundary condition. i.e. 1 to 20 are known to us.
1-57

Relaxation Technique (Iterative technique)

1-58

Relaxation Technique (Iterative technique)

1-59

Relaxation Technique (Iterative technique)

1-60

Relaxation Technique (Iterative technique)

1-61

Relaxation Technique (Iterative technique)

1-62

Relaxation Technique (Iterative technique)

1-63

Pressure correction technique

This technique is more suited to incompressible viscous flow
This technique further used in SIMPLE (semi implicit method for
pressure linked equation) for both compressible and incompressible
flow
Before describe the pressure correction technique, first we discussed
the two consideration associated with the incompressible flow solution

1-64

Incompressible Nevier-Stokes equation

The incompressible Nevier-Stokes equation will be derived from the
generalized Navier-Stokes equation by setting = constant hence

With further assumption =constant, reduce the Nevier-Stokes equation to,

1-65

Incompressible Nevier-Stokes equation

1-66