The aim of the thesis is to evaluate the effectiveness of stochastic models for particles distribution in complex turbulent flow. Turbulent two-phase flows exist in numerous natural phenomena and industrial systems, and remain a very challenging process to describe. Part of the difficulties originates from the fact that in most cases is not possible to provide an exact description for the continuous phase alone. Thus, a fundamental question arises: is it possible to obtain a reasonably accurate description of a dispersed system when simplified models are employed for the flow?

We consider a wall-mounted cube in a periodic turbulent channel flow with particles continuously injected after the rear face of the cube. The system is described through two alternative methodologies. First, accurate time-dependent numerical simulations are employed to provide reference data. In these simulations, the flow field is assumed to be exact and it directly and solely affects the particles dynamics. In the second approach, the flow is described via time-averaged numerical simulations which require additional ad hoc models introduced to describe the effect of turbulent fluctuations on the particles motion.