This thesis considers the prediction of turbulence in complex flows using
Reynolds Stress Model (RSM). The time evolution of the flow in the square ducts is simulated,
and the turbulence-driven secondary flow are predicted in the case of the square duct. Results are
compared to both experimental and CFD data. The mean secondary flow structures in the
square duct indicate the existence of strong, counter rotating vortex pairs, which are
symmetrically placed around the four outer corners of the inner square duct .The additive
multigrid method is used to accelerate the convergence of the pressure Poisson equation and the
appropriateness of this method to deal with the high wave number components in turbulence is
considered. Parallel computing techniques are applied to assist in the solution of the Navier-
This study is carried out using computational fluid dynamics (CFD) simulation
techniques as embedded in the commercially available CFD code (FLUENT 6.2). The CFD
modelling techniques solved the continuity, momentum and energy conservation equations.
Throughout the investigations, numerical validation is carried out by way of comparisons
of numerical results obtained from FLUENT to results reported in the work of other researcher.
Good agreement is found among both predictions.