The recent trend in energy research sought for a gas turbine combustor with highest possible thermal efficiency in compliance with the present environment regulation norms. Such a design improvement requires a detailed understanding of all the physical process involved from the air intake to turbine compressor to final exhaust.

The CFD is one of the most widely used technique for design and process optimization of gas turbine, that considerably reduces the cost involved and overall design time line.

The fuel injection is one of the vital process that determine the course of combustion inside the combustion chamber as it is primarily responsible for the fuel-air mixture formation and subsequent combustion.

The fuel injection itself is a complex process that is highly turbulence in nature and mean time it undergoes different kind of physical phenomena such as such as breakup (atomization), dispersion, evaporation and subsequent combustion.

The present work is mainly on the development and application of different mathematical sub-models to describe the physics of turbulent spray, which is typical for gas turbine combustors.

The applied models were formulated in the Eulerian-Lagrangian context in Lag-3D code and coupled with FASTEST code as a gas phase solver.

In order to quantify the instantaneous velocity seen by the particle as it appears in the particle motion equation and its effect on the particle distribution, dispersion model is needed. Turbulent dispersion is very important which influences the particle trajectories that are especially important when evaporation takes place. In the present thesis, three dispersion models, namely RWM-Iso, RWM-Aniso and PLM, are integrated and compared in this work.

Furthermore a systematical study in three configuration of different dispersion models and their influence on mass transfer and different turbulent intensities has been satisfactory carried out.

So the simulations have shown that the PLM model allows for achieving results that agree very well with the experimental measurement of the droplet mass flux in particular in upper sections. This means that the consideration of the advanced dispersion models like PLM enables to account well for anisotropic turbulence as well as vortex structures inherent to complex turbulent two phase flows.

In order to capture well the unsteady dynamics using RANS calculation, a modified SAS model has been adopted and compared with the standard k-epsilon model. Comparisons with experimental data shows that the SAS model is able to capture well the overall flow dynamics.

To represent the actual atomization dynamics for the dispersed phase in vicinity of particle injection nozzle, a stochastic model of drops air-blast breakup following the Kolmogorov´s model is implemented first time in Lag-3D and respective improvement is shown in the results.