You are on page 1of 1

a)Numerical methods tend to differ from analytical method because they relay on approximations

equation rather than exact equation.

Numerical methods use discretion which cause difference from exact solution due to round off
errors and simplification for physical modelling in numerical methods.

These differences can be improved by higher mesh densities, high grid quality, using high powered
computer and using exact model instead of simplified one.

b)

1) Residues
 Convergence criteria
2) Monitoring point
 Results not changing/ negligible difference in change
3) Mass, energy and scatter balance
 Reaction combustion
 Heat transfer radiation
 Multiple phase flow

c) Double precision solver is recommended to reduce round off errors, this allows computers to
store twice the number of digits reducing the roundoff errors.

D)

Finite volume method/ Finite difference method

Finite difference method Finite volume method


At each mesh node, the Taylor series Finite volume is divided the same way as FDM,
expansions are used to generate finite but instead relies on control volume, the
difference approximations to the partial control volume is a small volume surrounding
derivates of the equations. The derivates are each point on mesh. FVM solve by calculating
replaced by finite difference approximations, the values of the conserved variables averaged
yield algebraic equations for the flow solution across the volume.
at each mesh node point.

e)

Semi-Implicit Method for Pressure-Linked Equations( Simple)

Simple technique works for steady state problem by finding the solution iteratively. The solutions
use small iterations of equations with residuals, the iteration are stopped when the residual is
smaller the initial convergence criteria.

You might also like